Compare commits

..

17 Commits

Author SHA1 Message Date
antoniofilipovic
f5e3c26284 remove replication tests 2024-02-07 12:37:30 +01:00
antoniofilipovic
e6c8e7cfdc remove memory tests 2024-02-07 11:53:46 +01:00
antoniofilipovic
955609d036 Merge branch 'analyze-flakiness' of github.com:memgraph/memgraph into analyze-flakiness 2024-02-07 10:41:26 +01:00
antoniofilipovic
6c46046f3d remove schema test 2024-02-07 10:41:11 +01:00
antoniofilipovic
47db66c744 remove memory test, remove streams and lba on disk 2024-02-07 10:40:00 +01:00
antoniofilipovic
dcdeb19479 remove schema testing, remove memory limit testing 2024-02-06 17:32:03 +01:00
Deda
ff67e263cd Merge branch 'analyze-flakiness' of github.com:memgraph/memgraph into analyze-flakiness 2024-02-06 16:46:36 +01:00
Deda
4de5392aba Fix build type in analyze_e2e.yaml 2024-02-06 16:46:28 +01:00
antoniofilipovic
916af3006c Merge branch 'analyze-flakiness' of github.com:memgraph/memgraph into analyze-flakiness 2024-02-06 16:27:21 +01:00
antoniofilipovic
a7400e3b1d remove query modules tests 2024-02-06 16:27:09 +01:00
Marko Barišić
0e7c7c61ea Merge branch 'master' into analyze-flakiness 2024-02-05 13:09:05 +01:00
antoniofilipovic
35157a649e remove graphql tests 2024-02-05 10:37:00 +01:00
antoniofilipovic
4548baab0f Merge branch 'master' into analyze-flakiness 2024-02-05 10:34:36 +01:00
Deda
52383694f1 Add setup to analyze e2e tests 2024-01-31 16:38:55 +01:00
Deda
088773c0c8 Set fail-fast to false and max-parallel to 1 2024-01-30 11:24:33 +01:00
Deda
d6d36e4277 Add timeout and lower to 10 iterations 2024-01-30 00:54:18 +01:00
Deda
6b9389c545 Analyze core diff jobs for flakiness 2024-01-29 21:48:01 +01:00
251 changed files with 3144 additions and 7899 deletions

View File

@@ -6,7 +6,6 @@ Checks: '*,
-altera-unroll-loops,
-android-*,
-cert-err58-cpp,
-cppcoreguidelines-avoid-do-while,
-cppcoreguidelines-avoid-c-arrays,
-cppcoreguidelines-avoid-goto,
-cppcoreguidelines-avoid-magic-numbers,
@@ -61,7 +60,6 @@ Checks: '*,
-readability-implicit-bool-conversion,
-readability-magic-numbers,
-readability-named-parameter,
-readability-identifier-length,
-misc-no-recursion,
-concurrency-mt-unsafe,
-bugprone-easily-swappable-parameters'

649
.github/workflows/analyze_diff.yaml vendored Normal file
View File

@@ -0,0 +1,649 @@
name: Analyze diff flakiness
concurrency:
group: ${{ github.head_ref || github.sha }}
cancel-in-progress: true
on:
workflow_dispatch:
# pull_request:
# paths-ignore:
# - "docs/**"
# - "**/*.md"
# - ".clang-format"
# - "CODEOWNERS"
# - "licenses/*"
jobs:
community_build:
name: "Community build"
runs-on: [self-hosted, Linux, X64, Diff]
strategy:
max-parallel: 1
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 90
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Initialize deps
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
- name: Build community binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Build community binaries.
cd build
cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo -DMG_ENTERPRISE=OFF ..
make -j$THREADS
- name: Run unit tests
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Run unit tests.
cd build
ctest -R memgraph__unit --output-on-failure -j$THREADS
code_analysis:
name: "Code analysis"
runs-on: [self-hosted, Linux, X64, Diff]
strategy:
max-parallel: 1
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 90
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
# This is also needed if we want do to comparison against other branches
# See https://github.community/t/checkout-code-fails-when-it-runs-lerna-run-test-since-master/17920
- name: Fetch all history for all tags and branches
run: git fetch
- name: Initialize deps
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
- name: Set base branch
if: ${{ github.event_name == 'pull_request' }}
run: |
echo "BASE_BRANCH=origin/${{ github.base_ref }}" >> $GITHUB_ENV
- name: Set base branch # if we manually dispatch or push to master
if: ${{ github.event_name != 'pull_request' }}
run: |
echo "BASE_BRANCH=origin/master" >> $GITHUB_ENV
- name: Python code analysis
run: |
CHANGED_FILES=$(git diff -U0 ${{ env.BASE_BRANCH }}... --name-only --diff-filter=d)
for file in ${CHANGED_FILES}; do
echo ${file}
if [[ ${file} == *.py ]]; then
python3 -m black --check --diff ${file}
python3 -m isort --profile black --check-only --diff ${file}
fi
done
- name: Build combined ASAN, UBSAN and coverage binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
cd build
cmake -DTEST_COVERAGE=ON -DASAN=ON -DUBSAN=ON ..
make -j$THREADS memgraph__unit
- name: Run unit tests
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Run unit tests. It is restricted to 2 threads intentionally, because higher concurrency makes the timing related tests unstable.
cd build
LSAN_OPTIONS=suppressions=$PWD/../tools/lsan.supp UBSAN_OPTIONS=halt_on_error=1 ctest -R memgraph__unit --output-on-failure -j2
- name: Compute code coverage
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Compute code coverage.
cd tools/github
./coverage_convert
# Package code coverage.
cd generated
tar -czf code_coverage.tar.gz coverage.json html report.json summary.rmu
- name: Save code coverage
uses: actions/upload-artifact@v3
with:
name: "Code coverage"
path: tools/github/generated/code_coverage.tar.gz
- name: Run clang-tidy
run: |
source /opt/toolchain-v4/activate
# Restrict clang-tidy results only to the modified parts
git diff -U0 ${{ env.BASE_BRANCH }}... -- src | ./tools/github/clang-tidy/clang-tidy-diff.py -p 1 -j $THREADS -path build -regex ".+\.cpp" | tee ./build/clang_tidy_output.txt
# Fail if any warning is reported
! cat ./build/clang_tidy_output.txt | ./tools/github/clang-tidy/grep_error_lines.sh > /dev/null
debug_build:
name: "Debug build"
runs-on: [self-hosted, Linux, X64, Diff]
strategy:
max-parallel: 1
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 90
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Initialize deps
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
- name: Build debug binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Build debug binaries.
cd build
cmake ..
make -j$THREADS
- name: Run leftover CTest tests
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Run leftover CTest tests (all except unit and benchmark tests).
cd build
ctest -E "(memgraph__unit|memgraph__benchmark)" --output-on-failure
- name: Run drivers tests
run: |
./tests/drivers/run.sh
- name: Run integration tests
run: |
tests/integration/run.sh
- name: Run cppcheck and clang-format
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Run cppcheck and clang-format.
cd tools/github
./cppcheck_and_clang_format diff
- name: Save cppcheck and clang-format errors
uses: actions/upload-artifact@v3
with:
name: "Code coverage"
path: tools/github/cppcheck_and_clang_format.txt
release_build:
name: "Release build"
runs-on: [self-hosted, Linux, X64, Diff]
strategy:
max-parallel: 1
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 90
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Initialize deps
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
- name: Build release binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Build release binaries.
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make -j$THREADS
- name: Run GQL Behave tests
run: |
cd tests
./setup.sh /opt/toolchain-v4/activate
cd gql_behave
./continuous_integration
- name: Save quality assurance status
uses: actions/upload-artifact@v3
with:
name: "GQL Behave Status"
path: |
tests/gql_behave/gql_behave_status.csv
tests/gql_behave/gql_behave_status.html
- name: Run unit tests
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Run unit tests.
cd build
ctest -R memgraph__unit --output-on-failure -j$THREADS
- name: Ensure Kafka and Pulsar are up
run: |
cd tests/e2e/streams/kafka
docker-compose up -d
cd ../pulsar
docker-compose up -d
- name: Run e2e tests
run: |
cd tests
./setup.sh /opt/toolchain-v4/activate
source ve3/bin/activate_e2e
cd e2e
./run.sh
- name: Ensure Kafka and Pulsar are down
if: always()
run: |
cd tests/e2e/streams/kafka
docker-compose down
cd ../pulsar
docker-compose down
- name: Run stress test (plain)
run: |
cd tests/stress
source ve3/bin/activate
./continuous_integration
- name: Run stress test (SSL)
run: |
cd tests/stress
source ve3/bin/activate
./continuous_integration --use-ssl
- name: Run durability test
run: |
cd tests/stress
source ve3/bin/activate
python3 durability --num-steps 5
- name: Create enterprise DEB package
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
cd build
# create mgconsole
# we use the -B to force the build
make -j$THREADS -B mgconsole
# Create enterprise DEB package.
mkdir output && cd output
cpack -G DEB --config ../CPackConfig.cmake
- name: Save enterprise DEB package
uses: actions/upload-artifact@v3
with:
name: "Enterprise DEB package"
path: build/output/memgraph*.deb
- name: Save test data
uses: actions/upload-artifact@v3
if: always()
with:
name: "Test data"
path: |
# multiple paths could be defined
build/logs
#
# experimental_build_ha:
# name: "High availability build"
# runs-on: [self-hosted, Linux, X64, Diff]
# strategy:
# max-parallel: 1
# fail-fast: false
# matrix:
# try_id: [1,2,3,4,5,6,7,8,9,10]
# env:
# THREADS: 24
# MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
# MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
#
# steps:
# - name: Set up repository
# uses: actions/checkout@v3
# with:
# # Number of commits to fetch. `0` indicates all history for all
# # branches and tags. (default: 1)
# fetch-depth: 0
#
# - name: Initialize deps
# run: |
# # Activate toolchain.
# source /opt/toolchain-v4/activate
#
# # Initialize dependencies.
# ./init
#
# - name: Build release binaries
# run: |
# source /opt/toolchain-v4/activate
# cd build
# cmake -DCMAKE_BUILD_TYPE=Release -DMG_EXPERIMENTAL_HIGH_AVAILABILITY=ON ..
# make -j$THREADS
# - name: Run unit tests
# run: |
# source /opt/toolchain-v4/activate
# cd build
# ctest -R memgraph__unit --output-on-failure -j$THREADS
# - name: Run e2e tests
# run: |
# cd tests
# ./setup.sh /opt/toolchain-v4/activate
# source ve3/bin/activate_e2e
# cd e2e
# ./run.sh "Coordinator"
# ./run.sh "Client initiated failover"
# ./run.sh "Uninitialized cluster"
# - name: Save test data
# uses: actions/upload-artifact@v3
# if: always()
# with:
# name: "Test data"
# path: |
# # multiple paths could be defined
# build/logs
#
# experimental_build_mt:
# name: "MultiTenancy replication build"
# runs-on: [self-hosted, Linux, X64, Diff]
# strategy:
# max-parallel: 1
# fail-fast: false
# matrix:
# try_id: [1,2,3,4,5,6,7,8,9,10]
# env:
# THREADS: 24
# MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
# MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
#
# steps:
# - name: Set up repository
# uses: actions/checkout@v3
# with:
# # Number of commits to fetch. `0` indicates all history for all
# # branches and tags. (default: 1)
# fetch-depth: 0
#
#
# - name: Initialize deps
# run: |
# # Activate toolchain.
# source /opt/toolchain-v4/activate
#
# # Initialize dependencies.
# ./init
#
# - name: Build release binaries
# run: |
# # Activate toolchain.
# source /opt/toolchain-v4/activate
#
# # Build MT replication experimental binaries.
# cd build
# cmake -DCMAKE_BUILD_TYPE=Release -D MG_EXPERIMENTAL_REPLICATION_MULTITENANCY=ON ..
# make -j$THREADS
#
# - name: Run unit tests
# run: |
# # Activate toolchain.
# source /opt/toolchain-v4/activate
#
# # Run unit tests.
# cd build
# ctest -R memgraph__unit --output-on-failure -j$THREADS
#
# - name: Run e2e tests
# run: |
# cd tests
# ./setup.sh /opt/toolchain-v4/activate
# source ve3/bin/activate_e2e
# cd e2e
#
# # Just the replication based e2e tests
# ./run.sh "Replicate multitenancy"
# ./run.sh "Show"
# ./run.sh "Show while creating invalid state"
# ./run.sh "Delete edge replication"
# ./run.sh "Read-write benchmark"
# ./run.sh "Index replication"
# ./run.sh "Constraints"
#
# - name: Save test data
# uses: actions/upload-artifact@v3
# if: always()
# with:
# name: "Test data"
# path: |
# # multiple paths could be defined
# build/logs
release_jepsen_test:
name: "Release Jepsen Test"
runs-on: [self-hosted, Linux, X64, Debian10, JepsenControl]
strategy:
max-parallel: 1
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
#continue-on-error: true
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 90
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Initialize deps
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
- name: Build release binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Build only memgraph release binarie.
cd build
cmake -DCMAKE_BUILD_TYPE=RelWithDebInfo ..
make -j$THREADS memgraph
- name: Refresh Jepsen Cluster
run: |
cd tests/jepsen
./run.sh cluster-refresh
- name: Run Jepsen tests
run: |
cd tests/jepsen
./run.sh test-all-individually --binary ../../build/memgraph --ignore-run-stdout-logs --ignore-run-stderr-logs
- name: Save Jepsen report
uses: actions/upload-artifact@v3
if: ${{ always() }}
with:
name: "Jepsen Report"
path: tests/jepsen/Jepsen.tar.gz
release_benchmarks:
name: "Release benchmarks"
runs-on: [self-hosted, Linux, X64, Diff, Gen7]
strategy:
max-parallel: 1
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 90
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Initialize deps
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
- name: Build release binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Build only memgraph release binaries.
cd build
cmake -DCMAKE_BUILD_TYPE=release ..
make -j$THREADS
- name: Run macro benchmarks
run: |
cd tests/macro_benchmark
./harness QuerySuite MemgraphRunner \
--groups aggregation 1000_create unwind_create dense_expand match \
--no-strict
- name: Get branch name (merge)
if: github.event_name != 'pull_request'
shell: bash
run: echo "BRANCH_NAME=$(echo ${GITHUB_REF#refs/heads/} | tr / -)" >> $GITHUB_ENV
- name: Get branch name (pull request)
if: github.event_name == 'pull_request'
shell: bash
run: echo "BRANCH_NAME=$(echo ${GITHUB_HEAD_REF} | tr / -)" >> $GITHUB_ENV
- name: Upload macro benchmark results
run: |
cd tools/bench-graph-client
virtualenv -p python3 ve3
source ve3/bin/activate
pip install -r requirements.txt
./main.py --benchmark-name "macro_benchmark" \
--benchmark-results "../../tests/macro_benchmark/.harness_summary" \
--github-run-id "${{ github.run_id }}" \
--github-run-number "${{ github.run_number }}" \
--head-branch-name "${{ env.BRANCH_NAME }}"
# TODO (andi) No need for path flags and for --disk-storage and --in-memory-analytical
- name: Run mgbench
run: |
cd tests/mgbench
./benchmark.py vendor-native --num-workers-for-benchmark 12 --export-results benchmark_result.json pokec/medium/*/*
- name: Upload mgbench results
run: |
cd tools/bench-graph-client
virtualenv -p python3 ve3
source ve3/bin/activate
pip install -r requirements.txt
./main.py --benchmark-name "mgbench" \
--benchmark-results "../../tests/mgbench/benchmark_result.json" \
--github-run-id "${{ github.run_id }}" \
--github-run-number "${{ github.run_number }}" \
--head-branch-name "${{ env.BRANCH_NAME }}"

72
.github/workflows/analyze_e2e.yaml vendored Normal file
View File

@@ -0,0 +1,72 @@
name: Analyze e2e flakiness
concurrency:
group: ${{ github.head_ref || github.sha }}
cancel-in-progress: true
on:
workflow_dispatch:
pull_request:
paths-ignore:
- "docs/**"
- "**/*.md"
- ".clang-format"
- "CODEOWNERS"
- "licenses/*"
jobs:
release_e2e_test:
name: "Release End-to-end Test"
runs-on: [self-hosted, Linux, X64, Diff]
strategy:
max-parallel: 5
fail-fast: false
matrix:
try_id: [1,2,3,4,5,6,7,8,9,10]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
timeout-minutes: 60
steps:
- name: Set up repository
uses: actions/checkout@v3
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Build release binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
# Build release binaries
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make -j$THREADS
- name: Ensure Kafka and Pulsar are up
run: |
cd tests/e2e/streams/kafka
docker-compose up -d
cd ../pulsar
docker-compose up -d
- name: Run e2e tests
run: |
cd tests
./setup.sh /opt/toolchain-v4/activate
source ve3/bin/activate_e2e
cd e2e
./run.sh
- name: Ensure Kafka and Pulsar are down
if: always()
run: |
cd tests/e2e/streams/kafka
docker-compose down
cd ../pulsar
docker-compose down

View File

@@ -4,18 +4,18 @@ concurrency:
cancel-in-progress: true
on:
push:
branches:
- master
# push:
# branches:
# - master
workflow_dispatch:
pull_request:
paths-ignore:
- "docs/**"
- "**/*.md"
- ".clang-format"
- "CODEOWNERS"
- "licenses/*"
# pull_request:
# paths-ignore:
# - "docs/**"
# - "**/*.md"
# - ".clang-format"
# - "CODEOWNERS"
# - "licenses/*"
#
jobs:
community_build:
name: "Community build"
@@ -268,6 +268,7 @@ jobs:
ctest -R memgraph__unit --output-on-failure -j$THREADS
- name: Ensure Kafka and Pulsar are up
if: false
run: |
cd tests/e2e/streams/kafka
docker-compose up -d
@@ -275,6 +276,7 @@ jobs:
docker-compose up -d
- name: Run e2e tests
if: false
run: |
cd tests
./setup.sh /opt/toolchain-v4/activate
@@ -283,6 +285,7 @@ jobs:
./run.sh
- name: Ensure Kafka and Pulsar are down
if: false
run: |
cd tests/e2e/streams/kafka
docker-compose down
@@ -336,6 +339,118 @@ jobs:
# multiple paths could be defined
build/logs
experimental_build_ha:
name: "High availability build"
runs-on: [self-hosted, Linux, X64, Diff]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
steps:
- name: Set up repository
uses: actions/checkout@v4
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Build release binaries
run: |
source /opt/toolchain-v4/activate
./init
cd build
cmake -DCMAKE_BUILD_TYPE=Release -DMG_EXPERIMENTAL_HIGH_AVAILABILITY=ON ..
make -j$THREADS
- name: Run unit tests
run: |
source /opt/toolchain-v4/activate
cd build
ctest -R memgraph__unit --output-on-failure -j$THREADS
- name: Run e2e tests
if: false
run: |
cd tests
./setup.sh /opt/toolchain-v4/activate
source ve3/bin/activate_e2e
cd e2e
./run.sh "Coordinator"
./run.sh "Client initiated failover"
./run.sh "Uninitialized cluster"
- name: Save test data
uses: actions/upload-artifact@v4
if: always()
with:
name: "Test data(High availability build)"
path: |
# multiple paths could be defined
build/logs
experimental_build_mt:
name: "MultiTenancy replication build"
runs-on: [self-hosted, Linux, X64, Diff]
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
steps:
- name: Set up repository
uses: actions/checkout@v4
with:
# Number of commits to fetch. `0` indicates all history for all
# branches and tags. (default: 1)
fetch-depth: 0
- name: Build release binaries
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Initialize dependencies.
./init
# Build MT replication experimental binaries.
cd build
cmake -DCMAKE_BUILD_TYPE=Release -D MG_EXPERIMENTAL_REPLICATION_MULTITENANCY=ON ..
make -j$THREADS
- name: Run unit tests
run: |
# Activate toolchain.
source /opt/toolchain-v4/activate
# Run unit tests.
cd build
ctest -R memgraph__unit --output-on-failure -j$THREADS
- name: Run e2e tests
if: false
run: |
cd tests
./setup.sh /opt/toolchain-v4/activate
source ve3/bin/activate_e2e
cd e2e
# Just the replication based e2e tests
./run.sh "Replicate multitenancy"
./run.sh "Show"
./run.sh "Show while creating invalid state"
./run.sh "Delete edge replication"
./run.sh "Read-write benchmark"
./run.sh "Index replication"
./run.sh "Constraints"
- name: Save test data
uses: actions/upload-artifact@v4
if: always()
with:
name: "Test data(MultiTenancy replication build)"
path: |
# multiple paths could be defined
build/logs
release_jepsen_test:
name: "Release Jepsen Test"
runs-on: [self-hosted, Linux, X64, Debian10, JepsenControl]

View File

@@ -1,162 +0,0 @@
name: Release build test
concurrency:
group: ${{ github.workflow }}-${{ github.ref_name }}
cancel-in-progress: true
on:
workflow_dispatch:
inputs:
build_type:
type: choice
description: "Memgraph Build type. Default value is Release."
default: 'Release'
options:
- Release
- RelWithDebInfo
push:
branches:
- "release/**"
tags:
- "v*.*.*-rc*"
- "v*.*-rc*"
schedule:
# UTC
- cron: "0 22 * * *"
env:
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
BUILD_TYPE: ${{ github.event.inputs.build_type || 'Release' }}
jobs:
Debian10:
uses: ./.github/workflows/release_debian10.yaml
with:
build_type: ${{ github.event.inputs.build_type || 'Release' }}
secrets: inherit
Ubuntu20_04:
uses: ./.github/workflows/release_ubuntu2004.yaml
with:
build_type: ${{ github.event.inputs.build_type || 'Release' }}
secrets: inherit
PackageDebian10:
if: github.ref_type == 'tag'
needs: [Debian10]
runs-on: [self-hosted, DockerMgBuild, X64]
timeout-minutes: 60
steps:
- name: "Set up repository"
uses: actions/checkout@v4
with:
fetch-depth: 0 # Required because of release/get_version.py
- name: "Build package"
run: |
./release/package/run.sh package debian-10 $BUILD_TYPE
- name: "Upload package"
uses: actions/upload-artifact@v4
with:
name: debian-10
path: build/output/debian-10/memgraph*.deb
PackageUbuntu20_04:
if: github.ref_type == 'tag'
needs: [Ubuntu20_04]
runs-on: [self-hosted, DockerMgBuild, X64]
timeout-minutes: 60
steps:
- name: "Set up repository"
uses: actions/checkout@v4
with:
fetch-depth: 0 # Required because of release/get_version.py
- name: "Build package"
run: |
./release/package/run.sh package ubuntu-22.04 $BUILD_TYPE
- name: "Upload package"
uses: actions/upload-artifact@v4
with:
name: ubuntu-22.04
path: build/output/ubuntu-22.04/memgraph*.deb
PackageUbuntu20_04_ARM:
if: github.ref_type == 'tag'
needs: [Ubuntu20_04]
runs-on: [self-hosted, DockerMgBuild, ARM64]
# M1 Mac mini is sometimes slower
timeout-minutes: 90
steps:
- name: "Set up repository"
uses: actions/checkout@v4
with:
fetch-depth: 0 # Required because of release/get_version.py
- name: "Build package"
run: |
./release/package/run.sh package ubuntu-22.04-arm $BUILD_TYPE
- name: "Upload package"
uses: actions/upload-artifact@v4
with:
name: ubuntu-22.04-aarch64
path: build/output/ubuntu-22.04-arm/memgraph*.deb
PackageDebian11:
if: github.ref_type == 'tag'
needs: [Debian10, Ubuntu20_04]
runs-on: [self-hosted, DockerMgBuild, X64]
timeout-minutes: 60
steps:
- name: "Set up repository"
uses: actions/checkout@v4
with:
fetch-depth: 0 # Required because of release/get_version.py
- name: "Build package"
run: |
./release/package/run.sh package debian-11 $BUILD_TYPE
- name: "Upload package"
uses: actions/upload-artifact@v4
with:
name: debian-11
path: build/output/debian-11/memgraph*.deb
PackageDebian11_ARM:
if: github.ref_type == 'tag'
needs: [Debian10, Ubuntu20_04]
runs-on: [self-hosted, DockerMgBuild, ARM64]
# M1 Mac mini is sometimes slower
timeout-minutes: 90
steps:
- name: "Set up repository"
uses: actions/checkout@v4
with:
fetch-depth: 0 # Required because of release/get_version.py
- name: "Build package"
run: |
./release/package/run.sh package debian-11-arm $BUILD_TYPE
- name: "Upload package"
uses: actions/upload-artifact@v4
with:
name: debian-11-aarch64
path: build/output/debian-11-arm/memgraph*.deb
PushToS3:
if: github.ref_type == 'tag'
needs: [PackageDebian10, PackageDebian11, PackageDebian11_ARM, PackageUbuntu20_04, PackageUbuntu20_04_ARM]
runs-on: ubuntu-latest
steps:
- name: Download artifacts
uses: actions/download-artifact@v4
with:
# name: # if name input parameter is not provided, all artifacts are downloaded
# and put in directories named after each one.
path: build/output/release
- name: Upload to S3
uses: jakejarvis/s3-sync-action@v0.5.1
env:
AWS_S3_BUCKET: "deps.memgraph.io"
AWS_ACCESS_KEY_ID: ${{ secrets.S3_AWS_ACCESS_KEY_ID }}
AWS_SECRET_ACCESS_KEY: ${{ secrets.S3_AWS_SECRET_ACCESS_KEY }}
AWS_REGION: "eu-west-1"
SOURCE_DIR: "build/output/release"
DEST_DIR: "memgraph-unofficial/${{ github.ref_name }}/"

View File

@@ -1,12 +1,6 @@
name: Release Debian 10
on:
workflow_call:
inputs:
build_type:
type: string
description: "Memgraph Build type. Default value is Release."
default: 'Release'
workflow_dispatch:
inputs:
build_type:
@@ -17,8 +11,10 @@ on:
- Release
- RelWithDebInfo
schedule:
- cron: "0 22 * * *"
env:
OS: "Debian10"
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
@@ -115,7 +111,7 @@ jobs:
- name: Save code coverage
uses: actions/upload-artifact@v4
with:
name: "Code coverage(Coverage build)-${{ env.OS }}"
name: "Code coverage(Coverage build)"
path: tools/github/generated/code_coverage.tar.gz
debug_build:
@@ -169,7 +165,7 @@ jobs:
- name: Save cppcheck and clang-format errors
uses: actions/upload-artifact@v4
with:
name: "Code coverage(Debug build)-${{ env.OS }}"
name: "Code coverage(Debug build)"
path: tools/github/cppcheck_and_clang_format.txt
debug_integration_test:
@@ -246,7 +242,7 @@ jobs:
- name: Save enterprise DEB package
uses: actions/upload-artifact@v4
with:
name: "Enterprise DEB package-${{ env.OS}}"
name: "Enterprise DEB package"
path: build/output/memgraph*.deb
- name: Run GQL Behave tests
@@ -259,7 +255,7 @@ jobs:
- name: Save quality assurance status
uses: actions/upload-artifact@v4
with:
name: "GQL Behave Status-${{ env.OS }}"
name: "GQL Behave Status"
path: |
tests/gql_behave/gql_behave_status.csv
tests/gql_behave/gql_behave_status.html
@@ -325,6 +321,7 @@ jobs:
--no-strict
release_e2e_test:
if: false
name: "Release End-to-end Test"
runs-on: [self-hosted, Linux, X64, Debian10]
timeout-minutes: 60
@@ -459,5 +456,5 @@ jobs:
uses: actions/upload-artifact@v4
if: ${{ always() }}
with:
name: "Jepsen Report-${{ env.OS }}"
name: "Jepsen Report"
path: tests/jepsen/Jepsen.tar.gz

View File

@@ -1,12 +1,6 @@
name: Release Ubuntu 20.04
on:
workflow_call:
inputs:
build_type:
type: string
description: "Memgraph Build type. Default value is Release."
default: 'Release'
workflow_dispatch:
inputs:
build_type:
@@ -17,8 +11,10 @@ on:
- Release
- RelWithDebInfo
schedule:
- cron: "0 22 * * *"
env:
OS: "Ubuntu 20.04"
THREADS: 24
MEMGRAPH_ENTERPRISE_LICENSE: ${{ secrets.MEMGRAPH_ENTERPRISE_LICENSE }}
MEMGRAPH_ORGANIZATION_NAME: ${{ secrets.MEMGRAPH_ORGANIZATION_NAME }}
@@ -111,7 +107,7 @@ jobs:
- name: Save code coverage
uses: actions/upload-artifact@v4
with:
name: "Code coverage(Coverage build)-${{ env.OS }}"
name: "Code coverage(Coverage build)"
path: tools/github/generated/code_coverage.tar.gz
debug_build:
@@ -165,7 +161,7 @@ jobs:
- name: Save cppcheck and clang-format errors
uses: actions/upload-artifact@v4
with:
name: "Code coverage(Debug build)-${{ env.OS }}"
name: "Code coverage(Debug build)"
path: tools/github/cppcheck_and_clang_format.txt
debug_integration_test:
@@ -242,7 +238,7 @@ jobs:
- name: Save enterprise DEB package
uses: actions/upload-artifact@v4
with:
name: "Enterprise DEB package-${{ env.OS }}"
name: "Enterprise DEB package"
path: build/output/memgraph*.deb
- name: Run GQL Behave tests
@@ -255,7 +251,7 @@ jobs:
- name: Save quality assurance status
uses: actions/upload-artifact@v4
with:
name: "GQL Behave Status-${{ env.OS }}"
name: "GQL Behave Status"
path: |
tests/gql_behave/gql_behave_status.csv
tests/gql_behave/gql_behave_status.html
@@ -321,6 +317,7 @@ jobs:
--no-strict
release_e2e_test:
if: false
name: "Release End-to-end Test"
runs-on: [self-hosted, Linux, X64, Ubuntu20.04]
timeout-minutes: 60

View File

@@ -1,7 +1,4 @@
name: Stress test large
concurrency:
group: ${{ github.workflow }}-${{ github.ref_name }}
cancel-in-progress: true
on:
workflow_dispatch:
@@ -13,10 +10,7 @@ on:
options:
- Release
- RelWithDebInfo
push:
tags:
- "v*.*.*-rc*"
- "v*.*-rc*"
schedule:
- cron: "0 22 * * *"

View File

@@ -64,11 +64,11 @@ option(MG_ENTERPRISE "Build Memgraph Enterprise Edition" ON)
# Set the current version here to override the automatic version detection. The
# version must be specified as `X.Y.Z`. Primarily used when building new patch
# versions.
set(MEMGRAPH_OVERRIDE_VERSION "2.15.0")
set(MEMGRAPH_OVERRIDE_VERSION "")
# Custom suffix that this version should have. The suffix can be any arbitrary
# string. Primarily used when building a version for a specific customer.
set(MEMGRAPH_OVERRIDE_VERSION_SUFFIX "rc3")
set(MEMGRAPH_OVERRIDE_VERSION_SUFFIX "")
# Variables used to generate the versions.
if (MG_ENTERPRISE)
@@ -211,13 +211,8 @@ set(CMAKE_CXX_FLAGS_RELWITHDEBINFO
# ** Static linking is allowed only for executables! **
set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static-libgcc -static-libstdc++")
# Use lld linker to speedup build and use less memory.
add_link_options(-fuse-ld=lld)
# NOTE: Moving to latest Clang (probably starting from 15), lld stopped to work
# without explicit link_directories call.
string(REPLACE ":" " " LD_LIBS $ENV{LD_LIBRARY_PATH})
separate_arguments(LD_LIBS)
link_directories(${LD_LIBS})
# Use lld linker to speedup build
add_link_options(-fuse-ld=lld) # TODO: use mold linker
# release flags
set(CMAKE_CXX_FLAGS_RELEASE "-O2 -DNDEBUG")
@@ -276,6 +271,18 @@ endif()
set(libs_dir ${CMAKE_SOURCE_DIR}/libs)
add_subdirectory(libs EXCLUDE_FROM_ALL)
option(MG_EXPERIMENTAL_HIGH_AVAILABILITY "Feature flag for experimental high availability" OFF)
if (NOT MG_ENTERPRISE AND MG_EXPERIMENTAL_HIGH_AVAILABILITY)
set(MG_EXPERIMENTAL_HIGH_AVAILABILITY OFF)
message(FATAL_ERROR "MG_EXPERIMENTAL_HIGH_AVAILABILITY must be used with enterpise version of the code.")
endif ()
if (MG_EXPERIMENTAL_HIGH_AVAILABILITY)
add_compile_definitions(MG_EXPERIMENTAL_HIGH_AVAILABILITY)
endif ()
# Optional subproject configuration -------------------------------------------
option(TEST_COVERAGE "Generate coverage reports from running memgraph" OFF)
option(TOOLS "Build tools binaries" ON)
option(QUERY_MODULES "Build query modules containing custom procedures" ON)
@@ -284,6 +291,16 @@ option(TSAN "Build with Thread Sanitizer. To get a reasonable performance option
option(UBSAN "Build with Undefined Behaviour Sanitizer" OFF)
# Build feature flags
option(MG_EXPERIMENTAL_REPLICATION_MULTITENANCY "Feature flag for experimental replicaition of multitenacy" OFF)
if (NOT MG_ENTERPRISE AND MG_EXPERIMENTAL_REPLICATION_MULTITENANCY)
set(MG_EXPERIMENTAL_REPLICATION_MULTITENANCY OFF)
message(FATAL_ERROR "MG_EXPERIMENTAL_REPLICATION_MULTITENANCY with community edition build isn't possible")
endif ()
if (MG_EXPERIMENTAL_REPLICATION_MULTITENANCY)
add_compile_definitions(MG_EXPERIMENTAL_REPLICATION_MULTITENANCY)
endif ()
if (TEST_COVERAGE)
string(TOLOWER ${CMAKE_BUILD_TYPE} lower_build_type)

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -283,7 +283,7 @@ inline mgp_list *list_all_unique_constraints(mgp_graph *graph, mgp_memory *memor
}
// mgp_graph
inline bool graph_is_transactional(mgp_graph *graph) { return MgInvoke<int>(mgp_graph_is_transactional, graph); }
inline bool graph_is_mutable(mgp_graph *graph) { return MgInvoke<int>(mgp_graph_is_mutable, graph); }

1
libs/.gitignore vendored
View File

@@ -7,4 +7,3 @@
!pulsar.patch
!antlr4.10.1.patch
!rocksdb8.1.1.patch
!nuraft2.1.0.patch

View File

@@ -16,7 +16,7 @@ set(GFLAGS_NOTHREADS OFF)
# NOTE: config/generate.py depends on the gflags help XML format.
find_package(gflags REQUIRED)
find_package(fmt 8.0.1 REQUIRED)
find_package(fmt 8.0.1)
find_package(ZLIB 1.2.11 REQUIRED)
set(LIB_DIR ${CMAKE_CURRENT_SOURCE_DIR})

View File

@@ -5,7 +5,7 @@ index ee9b58c..31359a9 100644
@@ -48,7 +48,7 @@ option(LIBRDTSC_USE_PMU "Enables PMU usage on ARM platforms" OFF)
# | Library Build and Install Properties |
# +--------------------------------------------------------+
-add_library(rdtsc SHARED
+add_library(rdtsc
src/cycles.c
@@ -14,7 +14,7 @@ index ee9b58c..31359a9 100644
@@ -72,15 +72,6 @@ target_include_directories(rdtsc
PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include
)
-# Install directory changes depending on build mode
-if (CMAKE_BUILD_TYPE MATCHES "^[Dd]ebug")
- # During debug, the library will be installed into a local directory
@@ -27,15 +27,3 @@ index ee9b58c..31359a9 100644
# Specifying what to export when installing (GNUInstallDirs required)
install(TARGETS rdtsc
EXPORT librstsc-config
diff --git a/include/librdtsc/common_timer.h b/include/librdtsc/common_timer.h
index a6922d8..080dc77 100644
--- a/include/librdtsc/common_timer.h
+++ b/include/librdtsc/common_timer.h
@@ -2,6 +2,7 @@
#define LIBRDTSC_COMMON_TIMER_H
#include <librdtsc/common.h>
+#include <librdtsc/cycles.h>
extern uint64_t rdtsc_get_tsc_freq_arch();
extern uint64_t rdtsc_get_tsc_freq();

View File

@@ -1,24 +0,0 @@
diff --git a/include/libnuraft/asio_service_options.hxx b/include/libnuraft/asio_service_options.hxx
index 8fe1ec9..9497355 100644
--- a/include/libnuraft/asio_service_options.hxx
+++ b/include/libnuraft/asio_service_options.hxx
@@ -17,6 +17,7 @@ limitations under the License.
#pragma once
+#include <cstdint>
#include <functional>
#include <string>
#include <system_error>
diff --git a/include/libnuraft/callback.hxx b/include/libnuraft/callback.hxx
index 7b71624..d48c1e2 100644
--- a/include/libnuraft/callback.hxx
+++ b/include/libnuraft/callback.hxx
@@ -18,6 +18,7 @@ limitations under the License.
#ifndef _CALLBACK_H_
#define _CALLBACK_H_
+#include <cstdint>
#include <functional>
#include <string>

21
libs/rocksdb.patch Normal file
View File

@@ -0,0 +1,21 @@
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 6761929..6a369af 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -220,6 +220,7 @@ else()
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -momit-leaf-frame-pointer")
endif()
endif()
+ set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-deprecated-copy -Wno-unused-but-set-variable")
endif()
include(CheckCCompilerFlag)
@@ -997,7 +998,7 @@ if(NOT WIN32 OR ROCKSDB_INSTALL_ON_WINDOWS)
if(ROCKSDB_BUILD_SHARED)
install(
- TARGETS ${ROCKSDB_SHARED_LIB}
+ TARGETS ${ROCKSDB_SHARED_LIB} OPTIONAL
EXPORT RocksDBTargets
COMPONENT runtime
ARCHIVE DESTINATION "${CMAKE_INSTALL_LIBDIR}"

View File

@@ -123,10 +123,10 @@ declare -A primary_urls=(
["pulsar"]="http://$local_cache_host/git/pulsar.git"
["librdtsc"]="http://$local_cache_host/git/librdtsc.git"
["ctre"]="http://$local_cache_host/file/hanickadot/compile-time-regular-expressions/v3.7.2/single-header/ctre.hpp"
["absl"]="http://$local_cache_host/git/abseil-cpp.git"
["jemalloc"]="http://$local_cache_host/git/jemalloc.git"
["range-v3"]="http://$local_cache_host/git/range-v3.git"
["nuraft"]="http://$local_cache_host/git/NuRaft.git"
["absl"]="https://$local_cache_host/git/abseil-cpp.git"
["jemalloc"]="https://$local_cache_host/git/jemalloc.git"
["range-v3"]="https://$local_cache_host/git/ericniebler/range-v3.git"
["nuraft"]="https://$local_cache_host/git/eBay/NuRaft.git"
)
# The goal of secondary urls is to have links to the "source of truth" of
@@ -168,11 +168,12 @@ pushd antlr4
git apply ../antlr4.10.1.patch
popd
cppitertools_ref="v2.1" # 2021-01-15
# cppitertools v2.0 2019-12-23
cppitertools_ref="cb3635456bdb531121b82b4d2e3afc7ae1f56d47"
repo_clone_try_double "${primary_urls[cppitertools]}" "${secondary_urls[cppitertools]}" "cppitertools" "$cppitertools_ref"
# rapidcheck
rapidcheck_tag="1c91f40e64d87869250cfb610376c629307bf77d" # (2023-08-15)
rapidcheck_tag="7bc7d302191a4f3d0bf005692677126136e02f60" # (2020-05-04)
repo_clone_try_double "${primary_urls[rapidcheck]}" "${secondary_urls[rapidcheck]}" "rapidcheck" "$rapidcheck_tag"
# google benchmark
@@ -220,7 +221,7 @@ repo_clone_try_double "${primary_urls[pymgclient]}" "${secondary_urls[pymgclient
mgconsole_tag="v1.4.0" # (2023-05-21)
repo_clone_try_double "${primary_urls[mgconsole]}" "${secondary_urls[mgconsole]}" "mgconsole" "$mgconsole_tag" true
spdlog_tag="v1.12.0" # (2022-11-02)
spdlog_tag="v1.9.2" # (2021-08-12)
repo_clone_try_double "${primary_urls[spdlog]}" "${secondary_urls[spdlog]}" "spdlog" "$spdlog_tag" true
# librdkafka
@@ -269,8 +270,6 @@ pushd jemalloc
MALLOC_CONF="retain:false,percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:5000,dirty_decay_ms:5000" \
./configure \
--disable-cxx \
--with-lg-page=12 \
--with-lg-hugepage=21 \
--enable-shared=no --prefix=$working_dir \
--with-malloc-conf="retain:false,percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:5000,dirty_decay_ms:5000"
@@ -285,6 +284,5 @@ repo_clone_try_double "${primary_urls[range-v3]}" "${secondary_urls[range-v3]}"
nuraft_tag="v2.1.0"
repo_clone_try_double "${primary_urls[nuraft]}" "${secondary_urls[nuraft]}" "nuraft" "$nuraft_tag" true
pushd nuraft
git apply ../nuraft2.1.0.patch
./prepare.sh
popd

View File

@@ -36,7 +36,7 @@ ADDITIONAL USE GRANT: You may use the Licensed Work in accordance with the
3. using the Licensed Work to create a work or solution
which competes (or might reasonably be expected to
compete) with the Licensed Work.
CHANGE DATE: 2028-28-02
CHANGE DATE: 2028-21-01
CHANGE LICENSE: Apache License, Version 2.0
For information about alternative licensing arrangements, please visit: https://memgraph.com/legal.

View File

@@ -57,19 +57,16 @@ struct UpdateAuthData : memgraph::system::ISystemAction {
void DoDurability() override { /* Done during Auth execution */
}
bool DoReplication(replication::ReplicationClient &client, const utils::UUID &main_uuid,
replication::ReplicationEpoch const &epoch,
bool DoReplication(replication::ReplicationClient &client, replication::ReplicationEpoch const &epoch,
memgraph::system::Transaction const &txn) const override {
auto check_response = [](const replication::UpdateAuthDataRes &response) { return response.success; };
if (user_) {
return client.SteamAndFinalizeDelta<replication::UpdateAuthDataRpc>(
check_response, main_uuid, std::string{epoch.id()}, txn.last_committed_system_timestamp(), txn.timestamp(),
*user_);
check_response, std::string{epoch.id()}, txn.last_committed_system_timestamp(), txn.timestamp(), *user_);
}
if (role_) {
return client.SteamAndFinalizeDelta<replication::UpdateAuthDataRpc>(
check_response, main_uuid, std::string{epoch.id()}, txn.last_committed_system_timestamp(), txn.timestamp(),
*role_);
check_response, std::string{epoch.id()}, txn.last_committed_system_timestamp(), txn.timestamp(), *role_);
}
// Should never get here
MG_ASSERT(false, "Trying to update auth data that is not a user nor a role");
@@ -91,8 +88,7 @@ struct DropAuthData : memgraph::system::ISystemAction {
void DoDurability() override { /* Done during Auth execution */
}
bool DoReplication(replication::ReplicationClient &client, const utils::UUID &main_uuid,
replication::ReplicationEpoch const &epoch,
bool DoReplication(replication::ReplicationClient &client, replication::ReplicationEpoch const &epoch,
memgraph::system::Transaction const &txn) const override {
auto check_response = [](const replication::DropAuthDataRes &response) { return response.success; };
@@ -106,8 +102,7 @@ struct DropAuthData : memgraph::system::ISystemAction {
break;
}
return client.SteamAndFinalizeDelta<replication::DropAuthDataRpc>(
check_response, main_uuid, std::string{epoch.id()}, txn.last_committed_system_timestamp(), txn.timestamp(),
type, name_);
check_response, std::string{epoch.id()}, txn.last_committed_system_timestamp(), txn.timestamp(), type, name_);
}
void PostReplication(replication::RoleMainData &mainData) const override {}

View File

@@ -8,12 +8,10 @@
#pragma once
#include <cstdint>
#include <json/json.hpp>
#include <optional>
#include <string>
#include <json/json.hpp>
namespace memgraph::auth {
/// Need to be stable, auth durability depends on this
enum class PasswordHashAlgorithm : uint8_t { BCRYPT = 0, SHA256 = 1, SHA256_MULTIPLE = 2 };

View File

@@ -17,15 +17,8 @@
namespace memgraph::auth {
void LogWrongMain(const std::optional<utils::UUID> &current_main_uuid, const utils::UUID &main_req_id,
std::string_view rpc_req) {
spdlog::error(fmt::format("Received {} with main_id: {} != current_main_uuid: {}", rpc_req, std::string(main_req_id),
current_main_uuid.has_value() ? std::string(current_main_uuid.value()) : ""));
}
#ifdef MG_ENTERPRISE
void UpdateAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, auth::SynchedAuth &auth,
void UpdateAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access, auth::SynchedAuth &auth,
slk::Reader *req_reader, slk::Builder *res_builder) {
replication::UpdateAuthDataReq req;
memgraph::slk::Load(&req, req_reader);
@@ -33,12 +26,6 @@ void UpdateAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system
using memgraph::replication::UpdateAuthDataRes;
UpdateAuthDataRes res(false);
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, replication::UpdateAuthDataReq::kType.name);
memgraph::slk::Save(res, res_builder);
return;
}
// Note: No need to check epoch, recovery mechanism is done by a full uptodate snapshot
// of the set of databases. Hence no history exists to maintain regarding epoch change.
// If MAIN has changed we need to check this new group_timestamp is consistent with
@@ -66,8 +53,7 @@ void UpdateAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system
memgraph::slk::Save(res, res_builder);
}
void DropAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, auth::SynchedAuth &auth,
void DropAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access, auth::SynchedAuth &auth,
slk::Reader *req_reader, slk::Builder *res_builder) {
replication::DropAuthDataReq req;
memgraph::slk::Load(&req, req_reader);
@@ -75,12 +61,6 @@ void DropAuthDataHandler(memgraph::system::ReplicaHandlerAccessToState &system_s
using memgraph::replication::DropAuthDataRes;
DropAuthDataRes res(false);
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, replication::DropAuthDataRes::kType.name);
memgraph::slk::Save(res, res_builder);
return;
}
// Note: No need to check epoch, recovery mechanism is done by a full uptodate snapshot
// of the set of databases. Hence no history exists to maintain regarding epoch change.
// If MAIN has changed we need to check this new group_timestamp is consistent with
@@ -175,14 +155,14 @@ void Register(replication::RoleReplicaData const &data, system::ReplicaHandlerAc
auth::SynchedAuth &auth) {
// NOTE: Register even without license as the user could add a license at run-time
data.server->rpc_server_.Register<replication::UpdateAuthDataRpc>(
[&data, system_state_access, &auth](auto *req_reader, auto *res_builder) mutable {
[system_state_access, &auth](auto *req_reader, auto *res_builder) mutable {
spdlog::debug("Received UpdateAuthDataRpc");
UpdateAuthDataHandler(system_state_access, data.uuid_, auth, req_reader, res_builder);
UpdateAuthDataHandler(system_state_access, auth, req_reader, res_builder);
});
data.server->rpc_server_.Register<replication::DropAuthDataRpc>(
[&data, system_state_access, &auth](auto *req_reader, auto *res_builder) mutable {
[system_state_access, &auth](auto *req_reader, auto *res_builder) mutable {
spdlog::debug("Received DropAuthDataRpc");
DropAuthDataHandler(system_state_access, data.uuid_, auth, req_reader, res_builder);
DropAuthDataHandler(system_state_access, auth, req_reader, res_builder);
});
}
#endif

View File

@@ -17,16 +17,10 @@
#include "system/state.hpp"
namespace memgraph::auth {
void LogWrongMain(const std::optional<utils::UUID> &current_main_uuid, const utils::UUID &main_req_id,
std::string_view rpc_req);
#ifdef MG_ENTERPRISE
void UpdateAuthDataHandler(system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, auth::SynchedAuth &auth,
void UpdateAuthDataHandler(system::ReplicaHandlerAccessToState &system_state_access, auth::SynchedAuth &auth,
slk::Reader *req_reader, slk::Builder *res_builder);
void DropAuthDataHandler(system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, auth::SynchedAuth &auth,
void DropAuthDataHandler(system::ReplicaHandlerAccessToState &system_state_access, auth::SynchedAuth &auth,
slk::Reader *req_reader, slk::Builder *res_builder);
bool SystemRecoveryHandler(auth::SynchedAuth &auth, auth::Auth::Config auth_config,

View File

@@ -89,7 +89,6 @@ void Load(auth::Auth::Config *self, memgraph::slk::Reader *reader) {
// Serialize code for UpdateAuthDataReq
void Save(const memgraph::replication::UpdateAuthDataReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.main_uuid, builder);
memgraph::slk::Save(self.epoch_id, builder);
memgraph::slk::Save(self.expected_group_timestamp, builder);
memgraph::slk::Save(self.new_group_timestamp, builder);
@@ -97,7 +96,6 @@ void Save(const memgraph::replication::UpdateAuthDataReq &self, memgraph::slk::B
memgraph::slk::Save(self.role, builder);
}
void Load(memgraph::replication::UpdateAuthDataReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->main_uuid, reader);
memgraph::slk::Load(&self->epoch_id, reader);
memgraph::slk::Load(&self->expected_group_timestamp, reader);
memgraph::slk::Load(&self->new_group_timestamp, reader);
@@ -115,7 +113,6 @@ void Load(memgraph::replication::UpdateAuthDataRes *self, memgraph::slk::Reader
// Serialize code for DropAuthDataReq
void Save(const memgraph::replication::DropAuthDataReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.main_uuid, builder);
memgraph::slk::Save(self.epoch_id, builder);
memgraph::slk::Save(self.expected_group_timestamp, builder);
memgraph::slk::Save(self.new_group_timestamp, builder);
@@ -123,7 +120,6 @@ void Save(const memgraph::replication::DropAuthDataReq &self, memgraph::slk::Bui
memgraph::slk::Save(self.name, builder);
}
void Load(memgraph::replication::DropAuthDataReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->main_uuid, reader);
memgraph::slk::Load(&self->epoch_id, reader);
memgraph::slk::Load(&self->expected_group_timestamp, reader);
memgraph::slk::Load(&self->new_group_timestamp, reader);

View File

@@ -27,22 +27,17 @@ struct UpdateAuthDataReq {
static void Load(UpdateAuthDataReq *self, memgraph::slk::Reader *reader);
static void Save(const UpdateAuthDataReq &self, memgraph::slk::Builder *builder);
UpdateAuthDataReq() = default;
UpdateAuthDataReq(const utils::UUID &main_uuid, std::string epoch_id, uint64_t expected_ts, uint64_t new_ts,
auth::User user)
: main_uuid(main_uuid),
epoch_id{std::move(epoch_id)},
UpdateAuthDataReq(std::string epoch_id, uint64_t expected_ts, uint64_t new_ts, auth::User user)
: epoch_id{std::move(epoch_id)},
expected_group_timestamp{expected_ts},
new_group_timestamp{new_ts},
user{std::move(user)} {}
UpdateAuthDataReq(const utils::UUID &main_uuid, std::string epoch_id, uint64_t expected_ts, uint64_t new_ts,
auth::Role role)
: main_uuid(main_uuid),
epoch_id{std::move(epoch_id)},
UpdateAuthDataReq(std::string epoch_id, uint64_t expected_ts, uint64_t new_ts, auth::Role role)
: epoch_id{std::move(epoch_id)},
expected_group_timestamp{expected_ts},
new_group_timestamp{new_ts},
role{std::move(role)} {}
utils::UUID main_uuid;
std::string epoch_id;
uint64_t expected_group_timestamp;
uint64_t new_group_timestamp;
@@ -74,16 +69,13 @@ struct DropAuthDataReq {
enum class DataType { USER, ROLE };
DropAuthDataReq(const utils::UUID &main_uuid, std::string epoch_id, uint64_t expected_ts, uint64_t new_ts,
DataType type, std::string_view name)
: main_uuid(main_uuid),
epoch_id{std::move(epoch_id)},
DropAuthDataReq(std::string epoch_id, uint64_t expected_ts, uint64_t new_ts, DataType type, std::string_view name)
: epoch_id{std::move(epoch_id)},
expected_group_timestamp{expected_ts},
new_group_timestamp{new_ts},
type{type},
name{name} {}
utils::UUID main_uuid;
std::string epoch_id;
uint64_t expected_group_timestamp;
uint64_t new_group_timestamp;

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -15,9 +15,6 @@
#include "communication/bolt/v1/value.hpp"
#include "utils/logging.hpp"
#include "communication/bolt/v1/fmt.hpp"
#include "io/network/fmt.hpp"
namespace {
constexpr uint8_t kBoltV43Version[4] = {0x00, 0x00, 0x03, 0x04};
constexpr uint8_t kEmptyBoltVersion[4] = {0x00, 0x00, 0x00, 0x00};

View File

@@ -1,27 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#if FMT_VERSION > 90000
#include <fmt/ostream.h>
#include "communication/bolt/v1/value.hpp"
template <>
class fmt::formatter<memgraph::communication::bolt::Value> : public fmt::ostream_formatter {};
template <>
class fmt::formatter<std::vector<memgraph::communication::bolt::Value>> : public fmt::ostream_formatter {};
template <>
class fmt::formatter<std::map<std::string, memgraph::communication::bolt::Value>> : public fmt::ostream_formatter {};
#endif

View File

@@ -1,20 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#if FMT_VERSION > 90000
#include <fmt/ostream.h>
#include <boost/asio/ip/tcp.hpp>
template <>
class fmt::formatter<boost::asio::ip::tcp::endpoint> : public fmt::ostream_formatter {};
#endif

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -21,7 +21,6 @@
#include <boost/beast/core.hpp>
#include "communication/context.hpp"
#include "communication/fmt.hpp"
#include "communication/http/session.hpp"
#include "utils/spin_lock.hpp"
#include "utils/synchronized.hpp"
@@ -83,7 +82,7 @@ class Listener final : public std::enable_shared_from_this<Listener<TRequestHand
return;
}
spdlog::info("HTTP server is listening on {}", endpoint);
spdlog::info("HTTP server is listening on {}:{}", endpoint.address(), endpoint.port());
}
void DoAccept() {

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -23,7 +23,6 @@
#include "communication/session.hpp"
#include "io/network/epoll.hpp"
#include "io/network/fmt.hpp"
#include "io/network/socket.hpp"
#include "utils/logging.hpp"
#include "utils/signals.hpp"

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -22,7 +22,6 @@
#include "communication/init.hpp"
#include "communication/listener.hpp"
#include "io/network/fmt.hpp"
#include "io/network/socket.hpp"
#include "utils/logging.hpp"
#include "utils/message.hpp"

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -26,7 +26,6 @@
#include <boost/asio/ip/tcp.hpp>
#include "communication/context.hpp"
#include "communication/fmt.hpp"
#include "communication/init.hpp"
#include "communication/v2/listener.hpp"
#include "communication/v2/pool.hpp"
@@ -130,7 +129,7 @@ bool Server<TSession, TSessionContext>::Start() {
listener_->Start();
spdlog::info("{} server is fully armed and operational", service_name_);
spdlog::info("{} listening on {}", service_name_, endpoint_);
spdlog::info("{} listening on {}", service_name_, endpoint_.address());
context_thread_pool_.Run();
return true;

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -47,7 +47,6 @@
#include "communication/buffer.hpp"
#include "communication/context.hpp"
#include "communication/exceptions.hpp"
#include "communication/fmt.hpp"
#include "dbms/global.hpp"
#include "utils/event_counter.hpp"
#include "utils/logging.hpp"
@@ -213,11 +212,14 @@ class WebsocketSession : public std::enable_shared_from_this<WebsocketSession<TS
session_.Execute();
DoRead();
} catch (const SessionClosedException &e) {
spdlog::info("{} client {} closed the connection.", service_name_, remote_endpoint_);
spdlog::info("{} client {}:{} closed the connection.", service_name_, remote_endpoint_.address(),
remote_endpoint_.port());
DoClose();
} catch (const std::exception &e) {
spdlog::error("Exception was thrown while processing event in {} session associated with {}", service_name_,
remote_endpoint_);
spdlog::error(
"Exception was thrown while processing event in {} session "
"associated with {}:{}",
service_name_, remote_endpoint_.address(), remote_endpoint_.port());
spdlog::debug("Exception message: {}", e.what());
DoClose();
}
@@ -374,7 +376,8 @@ class Session final : public std::enable_shared_from_this<Session<TSession, TSes
socket.lowest_layer().non_blocking(false);
});
timeout_timer_.expires_at(boost::asio::steady_timer::time_point::max());
spdlog::info("Accepted a connection from {}: {}", service_name_, remote_endpoint_);
spdlog::info("Accepted a connection from {}: {}:{}", service_name_, remote_endpoint_.address(),
remote_endpoint_.port());
}
void DoRead() {
@@ -434,11 +437,14 @@ class Session final : public std::enable_shared_from_this<Session<TSession, TSes
session_.Execute();
DoRead();
} catch (const SessionClosedException &e) {
spdlog::info("{} client {} closed the connection.", service_name_, remote_endpoint_);
spdlog::info("{} client {}:{} closed the connection.", service_name_, remote_endpoint_.address(),
remote_endpoint_.port());
DoShutdown();
} catch (const std::exception &e) {
spdlog::error("Exception was thrown while processing event in {} session associated with {}", service_name_,
remote_endpoint_);
spdlog::error(
"Exception was thrown while processing event in {} session "
"associated with {}:{}",
service_name_, remote_endpoint_.address(), remote_endpoint_.port());
spdlog::debug("Exception message: {}", e.what());
DoShutdown();
}

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2022 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -10,7 +10,6 @@
// licenses/APL.txt.
#include "communication/websocket/listener.hpp"
#include "communication/fmt.hpp"
namespace memgraph::communication::websocket {
namespace {
@@ -62,7 +61,7 @@ Listener::Listener(boost::asio::io_context &ioc, ServerContext *context, tcp::en
return;
}
spdlog::info("WebSocket server is listening on {}", endpoint);
spdlog::info("WebSocket server is listening on {}:{}", endpoint.address(), endpoint.port());
}
void Listener::DoAccept() {

View File

@@ -8,31 +8,21 @@ target_sources(mg-coordination
include/coordination/coordinator_server.hpp
include/coordination/coordinator_config.hpp
include/coordination/coordinator_exceptions.hpp
include/coordination/coordinator_slk.hpp
include/coordination/coordinator_instance.hpp
include/coordination/coordinator_slk.hpp
include/coordination/coordinator_data.hpp
include/coordination/constants.hpp
include/coordination/coordinator_cluster_config.hpp
include/coordination/coordinator_handlers.hpp
include/coordination/instance_status.hpp
include/coordination/replication_instance.hpp
include/coordination/raft_state.hpp
include/coordination/rpc_errors.hpp
include/nuraft/coordinator_log_store.hpp
include/nuraft/coordinator_state_machine.hpp
include/nuraft/coordinator_state_manager.hpp
PRIVATE
coordinator_client.cpp
coordinator_state.cpp
coordinator_rpc.cpp
coordinator_server.cpp
coordinator_handlers.cpp
coordinator_data.cpp
coordinator_instance.cpp
replication_instance.cpp
raft_state.cpp
coordinator_log_store.cpp
coordinator_state_machine.cpp
coordinator_state_manager.cpp
coordinator_handlers.cpp
)
target_include_directories(mg-coordination PUBLIC include)

View File

@@ -9,7 +9,6 @@
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#include "utils/uuid.hpp"
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_client.hpp"
@@ -17,7 +16,6 @@
#include "coordination/coordinator_config.hpp"
#include "coordination/coordinator_rpc.hpp"
#include "replication_coordination_glue/messages.hpp"
#include "utils/result.hpp"
namespace memgraph::coordination {
@@ -29,38 +27,25 @@ auto CreateClientContext(memgraph::coordination::CoordinatorClientConfig const &
}
} // namespace
CoordinatorClient::CoordinatorClient(CoordinatorInstance *coord_instance, CoordinatorClientConfig config,
CoordinatorClient::CoordinatorClient(CoordinatorData *coord_data, CoordinatorClientConfig config,
HealthCheckCallback succ_cb, HealthCheckCallback fail_cb)
: rpc_context_{CreateClientContext(config)},
rpc_client_{io::network::Endpoint(io::network::Endpoint::needs_resolving, config.ip_address, config.port),
&rpc_context_},
config_{std::move(config)},
coord_instance_{coord_instance},
coord_data_{coord_data},
succ_cb_{std::move(succ_cb)},
fail_cb_{std::move(fail_cb)} {}
auto CoordinatorClient::InstanceName() const -> std::string { return config_.instance_name; }
auto CoordinatorClient::SocketAddress() const -> std::string { return rpc_client_.Endpoint().SocketAddress(); }
auto CoordinatorClient::InstanceDownTimeoutSec() const -> std::chrono::seconds {
return config_.instance_down_timeout_sec;
}
auto CoordinatorClient::InstanceGetUUIDFrequencySec() const -> std::chrono::seconds {
return config_.instance_get_uuid_frequency_sec;
}
void CoordinatorClient::StartFrequentCheck() {
if (instance_checker_.IsRunning()) {
return;
}
MG_ASSERT(config_.instance_health_check_frequency_sec > std::chrono::seconds(0),
MG_ASSERT(config_.health_check_frequency_sec > std::chrono::seconds(0),
"Health check frequency must be greater than 0");
instance_checker_.Run(
config_.instance_name, config_.instance_health_check_frequency_sec,
[this, instance_name = config_.instance_name] {
config_.instance_name, config_.health_check_frequency_sec, [this, instance_name = config_.instance_name] {
try {
spdlog::trace("Sending frequent heartbeat to machine {} on {}", instance_name,
rpc_client_.Endpoint().SocketAddress());
@@ -68,9 +53,9 @@ void CoordinatorClient::StartFrequentCheck() {
auto stream{rpc_client_.Stream<memgraph::replication_coordination_glue::FrequentHeartbeatRpc>()};
stream.AwaitResponse();
}
succ_cb_(coord_instance_, instance_name);
succ_cb_(coord_data_, instance_name);
} catch (rpc::RpcFailedException const &) {
fail_cb_(coord_instance_, instance_name);
fail_cb_(coord_data_, instance_name);
}
});
}
@@ -86,17 +71,16 @@ auto CoordinatorClient::SetCallbacks(HealthCheckCallback succ_cb, HealthCheckCal
auto CoordinatorClient::ReplicationClientInfo() const -> ReplClientInfo { return config_.replication_client_info; }
auto CoordinatorClient::SendPromoteReplicaToMainRpc(const utils::UUID &uuid,
ReplicationClientsInfo replication_clients_info) const -> bool {
auto CoordinatorClient::SendPromoteReplicaToMainRpc(ReplicationClientsInfo replication_clients_info) const -> bool {
try {
auto stream{rpc_client_.Stream<PromoteReplicaToMainRpc>(uuid, std::move(replication_clients_info))};
auto stream{rpc_client_.Stream<PromoteReplicaToMainRpc>(std::move(replication_clients_info))};
if (!stream.AwaitResponse().success) {
spdlog::error("Failed to receive successful PromoteReplicaToMainRpc response!");
spdlog::error("Failed to receive successful RPC failover response!");
return false;
}
return true;
} catch (rpc::RpcFailedException const &) {
spdlog::error("RPC error occurred while sending PromoteReplicaToMainRpc!");
spdlog::error("RPC error occurred while sending failover RPC!");
}
return false;
}
@@ -117,59 +101,5 @@ auto CoordinatorClient::DemoteToReplica() const -> bool {
return false;
}
auto CoordinatorClient::SendSwapMainUUIDRpc(const utils::UUID &uuid) const -> bool {
try {
auto stream{rpc_client_.Stream<replication_coordination_glue::SwapMainUUIDRpc>(uuid)};
if (!stream.AwaitResponse().success) {
spdlog::error("Failed to receive successful RPC swapping of uuid response!");
return false;
}
return true;
} catch (const rpc::RpcFailedException &) {
spdlog::error("RPC error occurred while sending swapping uuid RPC!");
}
return false;
}
auto CoordinatorClient::SendUnregisterReplicaRpc(std::string const &instance_name) const -> bool {
try {
auto stream{rpc_client_.Stream<UnregisterReplicaRpc>(instance_name)};
if (!stream.AwaitResponse().success) {
spdlog::error("Failed to receive successful RPC response for unregistering replica!");
return false;
}
return true;
} catch (rpc::RpcFailedException const &) {
spdlog::error("Failed to unregister replica!");
}
return false;
}
auto CoordinatorClient::SendGetInstanceUUIDRpc() const
-> utils::BasicResult<GetInstanceUUIDError, std::optional<utils::UUID>> {
try {
auto stream{rpc_client_.Stream<GetInstanceUUIDRpc>()};
auto res = stream.AwaitResponse();
return res.uuid;
} catch (const rpc::RpcFailedException &) {
spdlog::error("RPC error occured while sending GetInstance UUID RPC");
return GetInstanceUUIDError::RPC_EXCEPTION;
}
}
auto CoordinatorClient::SendEnableWritingOnMainRpc() const -> bool {
try {
auto stream{rpc_client_.Stream<EnableWritingOnMainRpc>()};
if (!stream.AwaitResponse().success) {
spdlog::error("Failed to receive successful RPC response for enabling writing on main!");
return false;
}
return true;
} catch (rpc::RpcFailedException const &) {
spdlog::error("Failed to enable writing on main!");
}
return false;
}
} // namespace memgraph::coordination
#endif

View File

@@ -0,0 +1,199 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#include "coordination/coordinator_instance.hpp"
#include "coordination/register_main_replica_coordinator_status.hpp"
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_data.hpp"
#include <range/v3/view.hpp>
#include <shared_mutex>
#include "libnuraft/nuraft.hxx"
namespace memgraph::coordination {
CoordinatorData::CoordinatorData() {
auto find_instance = [](CoordinatorData *coord_data, std::string_view instance_name) -> CoordinatorInstance & {
auto instance = std::ranges::find_if(
coord_data->registered_instances_,
[instance_name](CoordinatorInstance const &instance) { return instance.InstanceName() == instance_name; });
MG_ASSERT(instance != coord_data->registered_instances_.end(), "Instance {} not found during callback!",
instance_name);
return *instance;
};
replica_succ_cb_ = [find_instance](CoordinatorData *coord_data, std::string_view instance_name) -> void {
auto lock = std::lock_guard{coord_data->coord_data_lock_};
spdlog::trace("Instance {} performing replica successful callback", instance_name);
find_instance(coord_data, instance_name).OnSuccessPing();
};
replica_fail_cb_ = [find_instance](CoordinatorData *coord_data, std::string_view instance_name) -> void {
auto lock = std::lock_guard{coord_data->coord_data_lock_};
spdlog::trace("Instance {} performing replica failure callback", instance_name);
find_instance(coord_data, instance_name).OnFailPing();
};
main_succ_cb_ = [find_instance](CoordinatorData *coord_data, std::string_view instance_name) -> void {
auto lock = std::lock_guard{coord_data->coord_data_lock_};
spdlog::trace("Instance {} performing main successful callback", instance_name);
auto &instance = find_instance(coord_data, instance_name);
if (instance.IsAlive() || !coord_data->ClusterHasAliveMain_()) {
instance.OnSuccessPing();
return;
}
bool const demoted = instance.DemoteToReplica(coord_data->replica_succ_cb_, coord_data->replica_fail_cb_);
if (demoted) {
instance.OnSuccessPing();
spdlog::info("Instance {} demoted to replica", instance_name);
} else {
spdlog::error("Instance {} failed to become replica", instance_name);
}
};
main_fail_cb_ = [find_instance](CoordinatorData *coord_data, std::string_view instance_name) -> void {
auto lock = std::lock_guard{coord_data->coord_data_lock_};
spdlog::trace("Instance {} performing main failure callback", instance_name);
find_instance(coord_data, instance_name).OnFailPing();
if (!coord_data->ClusterHasAliveMain_()) {
spdlog::info("Cluster without main instance, trying automatic failover");
coord_data->TryFailover();
}
};
}
auto CoordinatorData::ClusterHasAliveMain_() const -> bool {
auto const alive_main = [](CoordinatorInstance const &instance) { return instance.IsMain() && instance.IsAlive(); };
return std::ranges::any_of(registered_instances_, alive_main);
}
auto CoordinatorData::TryFailover() -> void {
auto replica_instances = registered_instances_ | ranges::views::filter(&CoordinatorInstance::IsReplica);
auto chosen_replica_instance = std::ranges::find_if(replica_instances, &CoordinatorInstance::IsAlive);
if (chosen_replica_instance == replica_instances.end()) {
spdlog::warn("Failover failed since all replicas are down!");
return;
}
chosen_replica_instance->PauseFrequentCheck();
utils::OnScopeExit scope_exit{[&chosen_replica_instance] { chosen_replica_instance->ResumeFrequentCheck(); }};
std::vector<ReplClientInfo> repl_clients_info;
repl_clients_info.reserve(std::ranges::distance(replica_instances));
auto const not_chosen_replica_instance = [&chosen_replica_instance](CoordinatorInstance const &instance) {
return instance != *chosen_replica_instance;
};
std::ranges::transform(registered_instances_ | ranges::views::filter(not_chosen_replica_instance),
std::back_inserter(repl_clients_info),
[](const CoordinatorInstance &instance) { return instance.ReplicationClientInfo(); });
if (!chosen_replica_instance->PromoteToMain(std::move(repl_clients_info), main_succ_cb_, main_fail_cb_)) {
spdlog::warn("Failover failed since promoting replica to main failed!");
return;
}
spdlog::info("Failover successful! Instance {} promoted to main.", chosen_replica_instance->InstanceName());
}
auto CoordinatorData::ShowInstances() const -> std::vector<CoordinatorInstanceStatus> {
std::vector<CoordinatorInstanceStatus> instances_status;
instances_status.reserve(registered_instances_.size());
auto const stringify_repl_role = [](CoordinatorInstance const &instance) -> std::string {
if (!instance.IsAlive()) return "unknown";
if (instance.IsMain()) return "main";
return "replica";
};
auto const instance_to_status =
[&stringify_repl_role](CoordinatorInstance const &instance) -> CoordinatorInstanceStatus {
return {.instance_name = instance.InstanceName(),
.socket_address = instance.SocketAddress(),
.replication_role = stringify_repl_role(instance),
.is_alive = instance.IsAlive()};
};
{
auto lock = std::shared_lock{coord_data_lock_};
std::ranges::transform(registered_instances_, std::back_inserter(instances_status), instance_to_status);
}
return instances_status;
}
auto CoordinatorData::SetInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus {
auto lock = std::lock_guard{coord_data_lock_};
auto const is_new_main = [&instance_name](CoordinatorInstance const &instance) {
return instance.InstanceName() == instance_name;
};
auto new_main = std::ranges::find_if(registered_instances_, is_new_main);
if (new_main == registered_instances_.end()) {
spdlog::error("Instance {} not registered. Please register it using REGISTER INSTANCE {}", instance_name,
instance_name);
return SetInstanceToMainCoordinatorStatus::NO_INSTANCE_WITH_NAME;
}
new_main->PauseFrequentCheck();
utils::OnScopeExit scope_exit{[&new_main] { new_main->ResumeFrequentCheck(); }};
ReplicationClientsInfo repl_clients_info;
repl_clients_info.reserve(registered_instances_.size() - 1);
auto const is_not_new_main = [&instance_name](CoordinatorInstance const &instance) {
return instance.InstanceName() != instance_name;
};
std::ranges::transform(registered_instances_ | ranges::views::filter(is_not_new_main),
std::back_inserter(repl_clients_info),
[](const CoordinatorInstance &instance) { return instance.ReplicationClientInfo(); });
if (!new_main->PromoteToMain(std::move(repl_clients_info), main_succ_cb_, main_fail_cb_)) {
return SetInstanceToMainCoordinatorStatus::COULD_NOT_PROMOTE_TO_MAIN;
}
spdlog::info("Instance {} promoted to main", instance_name);
return SetInstanceToMainCoordinatorStatus::SUCCESS;
}
auto CoordinatorData::RegisterInstance(CoordinatorClientConfig config) -> RegisterInstanceCoordinatorStatus {
auto lock = std::lock_guard{coord_data_lock_};
if (std::ranges::any_of(registered_instances_, [&config](CoordinatorInstance const &instance) {
return instance.InstanceName() == config.instance_name;
})) {
return RegisterInstanceCoordinatorStatus::NAME_EXISTS;
}
if (std::ranges::any_of(registered_instances_, [&config](CoordinatorInstance const &instance) {
return instance.SocketAddress() == config.SocketAddress();
})) {
return RegisterInstanceCoordinatorStatus::ENDPOINT_EXISTS;
}
try {
registered_instances_.emplace_back(this, std::move(config), replica_succ_cb_, replica_fail_cb_);
return RegisterInstanceCoordinatorStatus::SUCCESS;
} catch (CoordinatorRegisterInstanceException const &) {
return RegisterInstanceCoordinatorStatus::RPC_FAILED;
}
}
} // namespace memgraph::coordination
#endif

View File

@@ -16,7 +16,6 @@
#include "coordination/coordinator_rpc.hpp"
#include "coordination/include/coordination/coordinator_server.hpp"
#include "replication/state.hpp"
namespace memgraph::dbms {
@@ -33,53 +32,18 @@ void CoordinatorHandlers::Register(memgraph::coordination::CoordinatorServer &se
spdlog::info("Received DemoteMainToReplicaRpc from coordinator server");
CoordinatorHandlers::DemoteMainToReplicaHandler(replication_handler, req_reader, res_builder);
});
server.Register<replication_coordination_glue::SwapMainUUIDRpc>(
[&replication_handler](slk::Reader *req_reader, slk::Builder *res_builder) -> void {
spdlog::info("Received SwapMainUUIDRPC on coordinator server");
CoordinatorHandlers::SwapMainUUIDHandler(replication_handler, req_reader, res_builder);
});
server.Register<coordination::UnregisterReplicaRpc>(
[&replication_handler](slk::Reader *req_reader, slk::Builder *res_builder) -> void {
spdlog::info("Received UnregisterReplicaRpc on coordinator server");
CoordinatorHandlers::UnregisterReplicaHandler(replication_handler, req_reader, res_builder);
});
server.Register<coordination::EnableWritingOnMainRpc>(
[&replication_handler](slk::Reader *req_reader, slk::Builder *res_builder) -> void {
spdlog::info("Received EnableWritingOnMainRpc on coordinator server");
CoordinatorHandlers::EnableWritingOnMainHandler(replication_handler, req_reader, res_builder);
});
server.Register<coordination::GetInstanceUUIDRpc>(
[&replication_handler](slk::Reader *req_reader, slk::Builder *res_builder) -> void {
spdlog::info("Received GetInstanceUUIDRpc on coordinator server");
CoordinatorHandlers::GetInstanceUUIDHandler(replication_handler, req_reader, res_builder);
});
}
void CoordinatorHandlers::SwapMainUUIDHandler(replication::ReplicationHandler &replication_handler,
slk::Reader *req_reader, slk::Builder *res_builder) {
if (!replication_handler.IsReplica()) {
spdlog::error("Setting main uuid must be performed on replica.");
slk::Save(replication_coordination_glue::SwapMainUUIDRes{false}, res_builder);
return;
}
replication_coordination_glue::SwapMainUUIDReq req;
slk::Load(&req, req_reader);
spdlog::info(fmt::format("Set replica data UUID to main uuid {}", std::string(req.uuid)));
std::get<memgraph::replication::RoleReplicaData>(replication_handler.GetReplState().ReplicationData()).uuid_ =
req.uuid;
slk::Save(replication_coordination_glue::SwapMainUUIDRes{true}, res_builder);
}
void CoordinatorHandlers::DemoteMainToReplicaHandler(replication::ReplicationHandler &replication_handler,
slk::Reader *req_reader, slk::Builder *res_builder) {
spdlog::info("Executing DemoteMainToReplicaHandler");
if (!replication_handler.IsMain()) {
spdlog::error("Setting to replica must be performed on main.");
slk::Save(coordination::DemoteMainToReplicaRes{false}, res_builder);
return;
}
coordination::DemoteMainToReplicaReq req;
slk::Load(&req, req_reader);
@@ -87,20 +51,13 @@ void CoordinatorHandlers::DemoteMainToReplicaHandler(replication::ReplicationHan
.ip_address = req.replication_client_info.replication_ip_address,
.port = req.replication_client_info.replication_port};
if (!replication_handler.SetReplicationRoleReplica(clients_config, std::nullopt)) {
if (!replication_handler.SetReplicationRoleReplica(clients_config)) {
spdlog::error("Demoting main to replica failed!");
slk::Save(coordination::DemoteMainToReplicaRes{false}, res_builder);
slk::Save(coordination::PromoteReplicaToMainRes{false}, res_builder);
return;
}
slk::Save(coordination::DemoteMainToReplicaRes{true}, res_builder);
}
void CoordinatorHandlers::GetInstanceUUIDHandler(replication::ReplicationHandler &replication_handler,
slk::Reader * /*req_reader*/, slk::Builder *res_builder) {
spdlog::info("Executing GetInstanceUUIDHandler");
slk::Save(coordination::GetInstanceUUIDRes{replication_handler.GetReplicaUUID()}, res_builder);
slk::Save(coordination::PromoteReplicaToMainRes{true}, res_builder);
}
void CoordinatorHandlers::PromoteReplicaToMainHandler(replication::ReplicationHandler &replication_handler,
@@ -110,17 +67,18 @@ void CoordinatorHandlers::PromoteReplicaToMainHandler(replication::ReplicationHa
slk::Save(coordination::PromoteReplicaToMainRes{false}, res_builder);
return;
}
coordination::PromoteReplicaToMainReq req;
slk::Load(&req, req_reader);
// This can fail because of disk. If it does, the cluster state could get inconsistent.
// We don't handle disk issues.
if (const bool success = replication_handler.DoReplicaToMainPromotion(req.main_uuid_); !success) {
if (!replication_handler.DoReplicaToMainPromotion()) {
spdlog::error("Promoting replica to main failed!");
slk::Save(coordination::PromoteReplicaToMainRes{false}, res_builder);
return;
}
coordination::PromoteReplicaToMainReq req;
slk::Load(&req, req_reader);
auto const converter = [](const auto &repl_info_config) {
return replication::ReplicationClientConfig{
.name = repl_info_config.instance_name,
@@ -151,68 +109,15 @@ void CoordinatorHandlers::PromoteReplicaToMainHandler(replication::ReplicationHa
spdlog::error("Registered replica could not be persisted!");
slk::Save(coordination::PromoteReplicaToMainRes{false}, res_builder);
return;
case memgraph::query::RegisterReplicaError::ERROR_ACCEPTING_MAIN:
spdlog::error("Replica didn't accept change of main!");
slk::Save(coordination::PromoteReplicaToMainRes{false}, res_builder);
return;
case memgraph::query::RegisterReplicaError::CONNECTION_FAILED:
// Connection failure is not a fatal error
break;
}
}
}
spdlog::info("Promote replica to main was success {}", std::string(req.main_uuid_));
slk::Save(coordination::PromoteReplicaToMainRes{true}, res_builder);
}
void CoordinatorHandlers::UnregisterReplicaHandler(replication::ReplicationHandler &replication_handler,
slk::Reader *req_reader, slk::Builder *res_builder) {
if (!replication_handler.IsMain()) {
spdlog::error("Unregistering replica must be performed on main.");
slk::Save(coordination::UnregisterReplicaRes{false}, res_builder);
return;
}
coordination::UnregisterReplicaReq req;
slk::Load(&req, req_reader);
auto res = replication_handler.UnregisterReplica(req.instance_name);
switch (res) {
using enum memgraph::query::UnregisterReplicaResult;
case SUCCESS:
slk::Save(coordination::UnregisterReplicaRes{true}, res_builder);
break;
case NOT_MAIN:
spdlog::error("Unregistering replica must be performed on main.");
slk::Save(coordination::UnregisterReplicaRes{false}, res_builder);
break;
case CAN_NOT_UNREGISTER:
spdlog::error("Could not unregister replica.");
slk::Save(coordination::UnregisterReplicaRes{false}, res_builder);
break;
case COULD_NOT_BE_PERSISTED:
spdlog::error("Could not persist replica unregistration.");
slk::Save(coordination::UnregisterReplicaRes{false}, res_builder);
break;
}
}
void CoordinatorHandlers::EnableWritingOnMainHandler(replication::ReplicationHandler &replication_handler,
slk::Reader * /*req_reader*/, slk::Builder *res_builder) {
if (!replication_handler.IsMain()) {
spdlog::error("Enable writing on main must be performed on main!");
slk::Save(coordination::EnableWritingOnMainRes{false}, res_builder);
return;
}
if (!replication_handler.GetReplState().EnableWritingOnMain()) {
spdlog::error("Enabling writing on main failed!");
slk::Save(coordination::EnableWritingOnMainRes{false}, res_builder);
return;
}
slk::Save(coordination::EnableWritingOnMainRes{true}, res_builder);
}
} // namespace memgraph::dbms
#endif

View File

@@ -13,335 +13,72 @@
#include "coordination/coordinator_instance.hpp"
#include "coordination/coordinator_exceptions.hpp"
#include "coordination/fmt.hpp"
#include "nuraft/coordinator_state_machine.hpp"
#include "nuraft/coordinator_state_manager.hpp"
#include "utils/counter.hpp"
#include "utils/functional.hpp"
#include <range/v3/view.hpp>
#include <shared_mutex>
namespace memgraph::coordination {
using nuraft::ptr;
using nuraft::srv_config;
CoordinatorInstance::CoordinatorInstance()
: raft_state_(RaftState::MakeRaftState(
[this] { std::ranges::for_each(repl_instances_, &ReplicationInstance::StartFrequentCheck); },
[this] { std::ranges::for_each(repl_instances_, &ReplicationInstance::StopFrequentCheck); })) {
auto find_repl_instance = [](CoordinatorInstance *self,
std::string_view repl_instance_name) -> ReplicationInstance & {
auto repl_instance =
std::ranges::find_if(self->repl_instances_, [repl_instance_name](ReplicationInstance const &instance) {
return instance.InstanceName() == repl_instance_name;
});
MG_ASSERT(repl_instance != self->repl_instances_.end(), "Instance {} not found during callback!",
repl_instance_name);
return *repl_instance;
};
replica_succ_cb_ = [find_repl_instance](CoordinatorInstance *self, std::string_view repl_instance_name) -> void {
auto lock = std::lock_guard{self->coord_instance_lock_};
spdlog::trace("Instance {} performing replica successful callback", repl_instance_name);
auto &repl_instance = find_repl_instance(self, repl_instance_name);
// We need to get replicas UUID from time to time to ensure replica is listening to correct main
// and that it didn't go down for less time than we could notice
// We need to get id of main replica is listening to
// and swap if necessary
if (!repl_instance.EnsureReplicaHasCorrectMainUUID(self->GetMainUUID())) {
spdlog::error("Failed to swap uuid for replica instance {} which is alive", repl_instance.InstanceName());
return;
}
repl_instance.OnSuccessPing();
};
replica_fail_cb_ = [find_repl_instance](CoordinatorInstance *self, std::string_view repl_instance_name) -> void {
auto lock = std::lock_guard{self->coord_instance_lock_};
spdlog::trace("Instance {} performing replica failure callback", repl_instance_name);
auto &repl_instance = find_repl_instance(self, repl_instance_name);
repl_instance.OnFailPing();
};
main_succ_cb_ = [find_repl_instance](CoordinatorInstance *self, std::string_view repl_instance_name) -> void {
auto lock = std::lock_guard{self->coord_instance_lock_};
spdlog::trace("Instance {} performing main successful callback", repl_instance_name);
auto &repl_instance = find_repl_instance(self, repl_instance_name);
if (repl_instance.IsAlive()) {
repl_instance.OnSuccessPing();
return;
}
const auto &repl_instance_uuid = repl_instance.GetMainUUID();
MG_ASSERT(repl_instance_uuid.has_value(), "Instance must have uuid set.");
auto const curr_main_uuid = self->GetMainUUID();
if (curr_main_uuid == repl_instance_uuid.value()) {
if (!repl_instance.EnableWritingOnMain()) {
spdlog::error("Failed to enable writing on main instance {}", repl_instance_name);
return;
}
repl_instance.OnSuccessPing();
return;
}
// TODO(antoniof) make demoteToReplica idempotent since main can be demoted to replica but
// swapUUID can fail
if (repl_instance.DemoteToReplica(self->replica_succ_cb_, self->replica_fail_cb_)) {
repl_instance.OnSuccessPing();
spdlog::info("Instance {} demoted to replica", repl_instance_name);
} else {
spdlog::error("Instance {} failed to become replica", repl_instance_name);
return;
}
if (!repl_instance.SendSwapAndUpdateUUID(curr_main_uuid)) {
spdlog::error(fmt::format("Failed to swap uuid for demoted main instance {}", repl_instance.InstanceName()));
return;
}
};
main_fail_cb_ = [find_repl_instance](CoordinatorInstance *self, std::string_view repl_instance_name) -> void {
auto lock = std::lock_guard{self->coord_instance_lock_};
spdlog::trace("Instance {} performing main failure callback", repl_instance_name);
auto &repl_instance = find_repl_instance(self, repl_instance_name);
repl_instance.OnFailPing();
const auto &repl_instance_uuid = repl_instance.GetMainUUID();
MG_ASSERT(repl_instance_uuid.has_value(), "Instance must have uuid set");
if (!repl_instance.IsAlive() && self->GetMainUUID() == repl_instance_uuid.value()) {
spdlog::info("Cluster without main instance, trying automatic failover");
self->TryFailover(); // TODO: (andi) Initiate failover
}
};
CoordinatorInstance::CoordinatorInstance(CoordinatorData *data, CoordinatorClientConfig config,
HealthCheckCallback succ_cb, HealthCheckCallback fail_cb)
: client_(data, std::move(config), std::move(succ_cb), std::move(fail_cb)),
replication_role_(replication_coordination_glue::ReplicationRole::REPLICA),
is_alive_(true) {
if (!client_.DemoteToReplica()) {
throw CoordinatorRegisterInstanceException("Failed to demote instance {} to replica", client_.InstanceName());
}
client_.StartFrequentCheck();
}
auto CoordinatorInstance::ShowInstances() const -> std::vector<InstanceStatus> {
auto const coord_instances = raft_state_.GetAllCoordinators();
auto const stringify_repl_role = [](ReplicationInstance const &instance) -> std::string {
if (!instance.IsAlive()) return "unknown";
if (instance.IsMain()) return "main";
return "replica";
};
auto const repl_instance_to_status = [&stringify_repl_role](ReplicationInstance const &instance) -> InstanceStatus {
return {.instance_name = instance.InstanceName(),
.coord_socket_address = instance.SocketAddress(),
.cluster_role = stringify_repl_role(instance),
.is_alive = instance.IsAlive()};
};
auto const coord_instance_to_status = [](ptr<srv_config> const &instance) -> InstanceStatus {
return {.instance_name = "coordinator_" + std::to_string(instance->get_id()),
.raft_socket_address = instance->get_endpoint(),
.cluster_role = "coordinator",
.is_alive = true}; // TODO: (andi) Get this info from RAFT and test it or when we will move
// CoordinatorState to every instance, we can be smarter about this using our RPC.
};
auto instances_status = utils::fmap(coord_instance_to_status, coord_instances);
{
auto lock = std::shared_lock{coord_instance_lock_};
std::ranges::transform(repl_instances_, std::back_inserter(instances_status), repl_instance_to_status);
}
return instances_status;
auto CoordinatorInstance::OnSuccessPing() -> void {
last_response_time_ = std::chrono::system_clock::now();
is_alive_ = true;
}
auto CoordinatorInstance::TryFailover() -> void {
auto alive_replicas = repl_instances_ | ranges::views::filter(&ReplicationInstance::IsReplica) |
ranges::views::filter(&ReplicationInstance::IsAlive);
if (ranges::empty(alive_replicas)) {
spdlog::warn("Failover failed since all replicas are down!");
return;
}
// TODO: Smarter choice
auto new_main = ranges::begin(alive_replicas);
new_main->PauseFrequentCheck();
utils::OnScopeExit scope_exit{[&new_main] { new_main->ResumeFrequentCheck(); }};
auto const is_not_new_main = [&new_main](ReplicationInstance &instance) {
return instance.InstanceName() != new_main->InstanceName();
};
auto const new_main_uuid = utils::UUID{};
// If for some replicas swap fails, for others on successful ping we will revert back on next change
// or we will do failover first again and then it will be consistent again
for (auto &other_replica_instance : alive_replicas | ranges::views::filter(is_not_new_main)) {
if (!other_replica_instance.SendSwapAndUpdateUUID(new_main_uuid)) {
spdlog::error(fmt::format("Failed to swap uuid for instance {} which is alive, aborting failover",
other_replica_instance.InstanceName()));
return;
}
}
auto repl_clients_info = repl_instances_ | ranges::views::filter(is_not_new_main) |
ranges::views::transform(&ReplicationInstance::ReplicationClientInfo) |
ranges::to<ReplicationClientsInfo>();
if (!new_main->PromoteToMain(new_main_uuid, std::move(repl_clients_info), main_succ_cb_, main_fail_cb_)) {
spdlog::warn("Failover failed since promoting replica to main failed!");
return;
}
// TODO: (andi) This should be replicated across all coordinator instances with Raft log
SetMainUUID(new_main_uuid);
spdlog::info("Failover successful! Instance {} promoted to main.", new_main->InstanceName());
auto CoordinatorInstance::OnFailPing() -> bool {
is_alive_ =
std::chrono::duration_cast<std::chrono::seconds>(std::chrono::system_clock::now() - last_response_time_).count() <
CoordinatorClusterConfig::alive_response_time_difference_sec_;
return is_alive_;
}
// TODO: (andi) Make sure you cannot put coordinator instance to the main
auto CoordinatorInstance::SetReplicationInstanceToMain(std::string instance_name)
-> SetInstanceToMainCoordinatorStatus {
auto lock = std::lock_guard{coord_instance_lock_};
auto CoordinatorInstance::InstanceName() const -> std::string { return client_.InstanceName(); }
auto CoordinatorInstance::SocketAddress() const -> std::string { return client_.SocketAddress(); }
auto CoordinatorInstance::IsAlive() const -> bool { return is_alive_; }
if (std::ranges::any_of(repl_instances_, &ReplicationInstance::IsMain)) {
return SetInstanceToMainCoordinatorStatus::MAIN_ALREADY_EXISTS;
}
auto const is_new_main = [&instance_name](ReplicationInstance const &instance) {
return instance.InstanceName() == instance_name;
};
auto new_main = std::ranges::find_if(repl_instances_, is_new_main);
if (new_main == repl_instances_.end()) {
spdlog::error("Instance {} not registered. Please register it using REGISTER INSTANCE {}", instance_name,
instance_name);
return SetInstanceToMainCoordinatorStatus::NO_INSTANCE_WITH_NAME;
}
new_main->PauseFrequentCheck();
utils::OnScopeExit scope_exit{[&new_main] { new_main->ResumeFrequentCheck(); }};
auto const is_not_new_main = [&instance_name](ReplicationInstance const &instance) {
return instance.InstanceName() != instance_name;
};
auto const new_main_uuid = utils::UUID{};
for (auto &other_instance : repl_instances_ | ranges::views::filter(is_not_new_main)) {
if (!other_instance.SendSwapAndUpdateUUID(new_main_uuid)) {
spdlog::error(
fmt::format("Failed to swap uuid for instance {}, aborting failover", other_instance.InstanceName()));
return SetInstanceToMainCoordinatorStatus::SWAP_UUID_FAILED;
}
}
ReplicationClientsInfo repl_clients_info;
repl_clients_info.reserve(repl_instances_.size() - 1);
std::ranges::transform(repl_instances_ | ranges::views::filter(is_not_new_main),
std::back_inserter(repl_clients_info), &ReplicationInstance::ReplicationClientInfo);
if (!new_main->PromoteToMain(new_main_uuid, std::move(repl_clients_info), main_succ_cb_, main_fail_cb_)) {
return SetInstanceToMainCoordinatorStatus::COULD_NOT_PROMOTE_TO_MAIN;
}
// TODO: (andi) This should be replicated across all coordinator instances with Raft log
SetMainUUID(new_main_uuid);
spdlog::info("Instance {} promoted to main", instance_name);
return SetInstanceToMainCoordinatorStatus::SUCCESS;
auto CoordinatorInstance::IsReplica() const -> bool {
return replication_role_ == replication_coordination_glue::ReplicationRole::REPLICA;
}
auto CoordinatorInstance::IsMain() const -> bool {
return replication_role_ == replication_coordination_glue::ReplicationRole::MAIN;
}
auto CoordinatorInstance::RegisterReplicationInstance(CoordinatorClientConfig config)
-> RegisterInstanceCoordinatorStatus {
auto lock = std::lock_guard{coord_instance_lock_};
auto instance_name = config.instance_name;
auto const name_matches = [&instance_name](ReplicationInstance const &instance) {
return instance.InstanceName() == instance_name;
};
if (std::ranges::any_of(repl_instances_, name_matches)) {
return RegisterInstanceCoordinatorStatus::NAME_EXISTS;
auto CoordinatorInstance::PromoteToMain(ReplicationClientsInfo repl_clients_info, HealthCheckCallback main_succ_cb,
HealthCheckCallback main_fail_cb) -> bool {
if (!client_.SendPromoteReplicaToMainRpc(std::move(repl_clients_info))) {
return false;
}
auto const socket_address_matches = [&config](ReplicationInstance const &instance) {
return instance.SocketAddress() == config.SocketAddress();
};
replication_role_ = replication_coordination_glue::ReplicationRole::MAIN;
client_.SetCallbacks(std::move(main_succ_cb), std::move(main_fail_cb));
if (std::ranges::any_of(repl_instances_, socket_address_matches)) {
return RegisterInstanceCoordinatorStatus::ENDPOINT_EXISTS;
}
if (!raft_state_.RequestLeadership()) {
return RegisterInstanceCoordinatorStatus::NOT_LEADER;
}
auto const res = raft_state_.AppendRegisterReplicationInstance(instance_name);
if (!res->get_accepted()) {
spdlog::error(
"Failed to accept request for registering instance {}. Most likely the reason is that the instance is not "
"the "
"leader.",
config.instance_name);
return RegisterInstanceCoordinatorStatus::RAFT_COULD_NOT_ACCEPT;
}
spdlog::info("Request for registering instance {} accepted", instance_name);
try {
repl_instances_.emplace_back(this, std::move(config), replica_succ_cb_, replica_fail_cb_);
} catch (CoordinatorRegisterInstanceException const &) {
return RegisterInstanceCoordinatorStatus::RPC_FAILED;
}
if (res->get_result_code() != nuraft::cmd_result_code::OK) {
spdlog::error("Failed to register instance {} with error code {}", instance_name, res->get_result_code());
return RegisterInstanceCoordinatorStatus::RAFT_COULD_NOT_APPEND;
}
spdlog::info("Instance {} registered", instance_name);
return RegisterInstanceCoordinatorStatus::SUCCESS;
return true;
}
auto CoordinatorInstance::UnregisterReplicationInstance(std::string instance_name)
-> UnregisterInstanceCoordinatorStatus {
auto lock = std::lock_guard{coord_instance_lock_};
auto const name_matches = [&instance_name](ReplicationInstance const &instance) {
return instance.InstanceName() == instance_name;
};
auto inst_to_remove = std::ranges::find_if(repl_instances_, name_matches);
if (inst_to_remove == repl_instances_.end()) {
return UnregisterInstanceCoordinatorStatus::NO_INSTANCE_WITH_NAME;
auto CoordinatorInstance::DemoteToReplica(HealthCheckCallback replica_succ_cb, HealthCheckCallback replica_fail_cb)
-> bool {
if (!client_.DemoteToReplica()) {
return false;
}
if (inst_to_remove->IsMain() && inst_to_remove->IsAlive()) {
return UnregisterInstanceCoordinatorStatus::IS_MAIN;
}
replication_role_ = replication_coordination_glue::ReplicationRole::REPLICA;
client_.SetCallbacks(std::move(replica_succ_cb), std::move(replica_fail_cb));
inst_to_remove->StopFrequentCheck();
auto curr_main = std::ranges::find_if(repl_instances_, &ReplicationInstance::IsMain);
MG_ASSERT(curr_main != repl_instances_.end(), "There must be a main instance when unregistering a replica");
if (!curr_main->SendUnregisterReplicaRpc(instance_name)) {
inst_to_remove->StartFrequentCheck();
return UnregisterInstanceCoordinatorStatus::RPC_FAILED;
}
std::erase_if(repl_instances_, name_matches);
return UnregisterInstanceCoordinatorStatus::SUCCESS;
return true;
}
auto CoordinatorInstance::AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address)
-> void {
raft_state_.AddCoordinatorInstance(raft_server_id, raft_port, std::move(raft_address));
auto CoordinatorInstance::PauseFrequentCheck() -> void { client_.PauseFrequentCheck(); }
auto CoordinatorInstance::ResumeFrequentCheck() -> void { client_.ResumeFrequentCheck(); }
auto CoordinatorInstance::ReplicationClientInfo() const -> CoordinatorClientConfig::ReplicationClientInfo {
return client_.ReplicationClientInfo();
}
auto CoordinatorInstance::GetMainUUID() const -> utils::UUID { return main_uuid_; }
// TODO: (andi) Add to the RAFT log.
auto CoordinatorInstance::SetMainUUID(utils::UUID new_uuid) -> void { main_uuid_ = new_uuid; }
} // namespace memgraph::coordination
#endif

View File

@@ -1,201 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#ifdef MG_ENTERPRISE
#include "nuraft/coordinator_log_store.hpp"
#include "coordination/coordinator_exceptions.hpp"
#include "utils/logging.hpp"
namespace memgraph::coordination {
using nuraft::cs_new;
namespace {
ptr<log_entry> MakeClone(const ptr<log_entry> &entry) {
return cs_new<log_entry>(entry->get_term(), buffer::clone(entry->get_buf()), entry->get_val_type(),
entry->get_timestamp());
}
} // namespace
CoordinatorLogStore::CoordinatorLogStore() : start_idx_(1) {
ptr<buffer> buf = buffer::alloc(sizeof(uint64_t));
logs_[0] = cs_new<log_entry>(0, buf);
}
CoordinatorLogStore::~CoordinatorLogStore() {}
auto CoordinatorLogStore::FindOrDefault_(uint64_t index) const -> ptr<log_entry> {
auto entry = logs_.find(index);
if (entry == logs_.end()) {
entry = logs_.find(0);
}
return entry->second;
}
uint64_t CoordinatorLogStore::next_slot() const {
auto lock = std::lock_guard{logs_lock_};
return start_idx_ + logs_.size() - 1;
}
uint64_t CoordinatorLogStore::start_index() const { return start_idx_; }
ptr<log_entry> CoordinatorLogStore::last_entry() const {
auto lock = std::lock_guard{logs_lock_};
uint64_t const last_idx = start_idx_ + logs_.size() - 1;
auto const last_src = FindOrDefault_(last_idx - 1);
return MakeClone(last_src);
}
uint64_t CoordinatorLogStore::append(ptr<log_entry> &entry) {
ptr<log_entry> clone = MakeClone(entry);
uint64_t next_slot{0};
{
auto lock = std::lock_guard{logs_lock_};
next_slot = start_idx_ + logs_.size() - 1;
logs_[next_slot] = clone;
}
return next_slot;
}
void CoordinatorLogStore::write_at(uint64_t index, ptr<log_entry> &entry) {
ptr<log_entry> clone = MakeClone(entry);
// Discard all logs equal to or greater than `index.
{
auto lock = std::lock_guard{logs_lock_};
auto itr = logs_.lower_bound(index);
while (itr != logs_.end()) {
itr = logs_.erase(itr);
}
logs_[index] = clone;
}
}
ptr<std::vector<ptr<log_entry>>> CoordinatorLogStore::log_entries(uint64_t start, uint64_t end) {
auto ret = cs_new<std::vector<ptr<log_entry>>>();
ret->resize(end - start);
for (uint64_t i = start, curr_index = 0; i < end; ++i, ++curr_index) {
ptr<log_entry> src = nullptr;
{
auto lock = std::lock_guard{logs_lock_};
if (auto const entry = logs_.find(i); entry != logs_.end()) {
src = entry->second;
} else {
throw RaftCouldNotFindEntryException("Could not find entry at index {}", i);
}
}
(*ret)[curr_index] = MakeClone(src);
}
return ret;
}
ptr<log_entry> CoordinatorLogStore::entry_at(uint64_t index) {
ptr<log_entry> src = nullptr;
{
auto lock = std::lock_guard{logs_lock_};
src = FindOrDefault_(index);
}
return MakeClone(src);
}
uint64_t CoordinatorLogStore::term_at(uint64_t index) {
uint64_t term = 0;
{
auto lock = std::lock_guard{logs_lock_};
term = FindOrDefault_(index)->get_term();
}
return term;
}
ptr<buffer> CoordinatorLogStore::pack(uint64_t index, int32 cnt) {
std::vector<ptr<buffer>> logs;
size_t size_total = 0;
uint64_t const end_index = index + cnt;
for (uint64_t i = index; i < end_index; ++i) {
ptr<log_entry> le = nullptr;
{
auto lock = std::lock_guard{logs_lock_};
le = logs_[i];
}
MG_ASSERT(le.get(), "Could not find log entry at index {}", i);
auto buf = le->serialize();
size_total += buf->size();
logs.push_back(buf);
}
auto buf_out = buffer::alloc(sizeof(int32) + cnt * sizeof(int32) + size_total);
buf_out->pos(0);
buf_out->put((int32)cnt);
for (auto &entry : logs) {
buf_out->put(static_cast<int32>(entry->size()));
buf_out->put(*entry);
}
return buf_out;
}
void CoordinatorLogStore::apply_pack(uint64_t index, buffer &pack) {
pack.pos(0);
int32 const num_logs = pack.get_int();
for (int32 i = 0; i < num_logs; ++i) {
uint64_t cur_idx = index + i;
int32 buf_size = pack.get_int();
ptr<buffer> buf_local = buffer::alloc(buf_size);
pack.get(buf_local);
ptr<log_entry> le = log_entry::deserialize(*buf_local);
{
auto lock = std::lock_guard{logs_lock_};
logs_[cur_idx] = le;
}
}
{
auto lock = std::lock_guard{logs_lock_};
auto const entry = logs_.upper_bound(0);
if (entry != logs_.end()) {
start_idx_ = entry->first;
} else {
start_idx_ = 1;
}
}
}
// NOTE: Remove all logs up to given 'last_log_index' (inclusive).
bool CoordinatorLogStore::compact(uint64_t last_log_index) {
auto lock = std::lock_guard{logs_lock_};
for (uint64_t ii = start_idx_; ii <= last_log_index; ++ii) {
auto const entry = logs_.find(ii);
if (entry != logs_.end()) {
logs_.erase(entry);
}
}
if (start_idx_ <= last_log_index) {
start_idx_ = last_log_index + 1;
}
return true;
}
bool CoordinatorLogStore::flush() { return true; }
} // namespace memgraph::coordination
#endif

View File

@@ -52,51 +52,6 @@ void DemoteMainToReplicaRes::Load(DemoteMainToReplicaRes *self, memgraph::slk::R
memgraph::slk::Load(self, reader);
}
void UnregisterReplicaReq::Save(UnregisterReplicaReq const &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self, builder);
}
void UnregisterReplicaReq::Load(UnregisterReplicaReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(self, reader);
}
void UnregisterReplicaRes::Save(UnregisterReplicaRes const &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self, builder);
}
void UnregisterReplicaRes::Load(UnregisterReplicaRes *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(self, reader);
}
void EnableWritingOnMainRes::Save(EnableWritingOnMainRes const &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self, builder);
}
void EnableWritingOnMainRes::Load(EnableWritingOnMainRes *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(self, reader);
}
void EnableWritingOnMainReq::Save(EnableWritingOnMainReq const &self, memgraph::slk::Builder *builder) {}
void EnableWritingOnMainReq::Load(EnableWritingOnMainReq *self, memgraph::slk::Reader *reader) {}
// GetInstanceUUID
void GetInstanceUUIDReq::Save(const GetInstanceUUIDReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self, builder);
}
void GetInstanceUUIDReq::Load(GetInstanceUUIDReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(self, reader);
}
void GetInstanceUUIDRes::Save(const GetInstanceUUIDRes &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self, builder);
}
void GetInstanceUUIDRes::Load(GetInstanceUUIDRes *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(self, reader);
}
} // namespace coordination
constexpr utils::TypeInfo coordination::PromoteReplicaToMainReq::kType{utils::TypeId::COORD_FAILOVER_REQ,
@@ -109,31 +64,10 @@ constexpr utils::TypeInfo coordination::DemoteMainToReplicaReq::kType{utils::Typ
"CoordDemoteToReplicaReq", nullptr};
constexpr utils::TypeInfo coordination::DemoteMainToReplicaRes::kType{utils::TypeId::COORD_SET_REPL_MAIN_RES,
"CoordDemoteToReplicaRes", nullptr};
constexpr utils::TypeInfo coordination::UnregisterReplicaReq::kType{utils::TypeId::COORD_UNREGISTER_REPLICA_REQ,
"UnregisterReplicaReq", nullptr};
constexpr utils::TypeInfo coordination::UnregisterReplicaRes::kType{utils::TypeId::COORD_UNREGISTER_REPLICA_RES,
"UnregisterReplicaRes", nullptr};
constexpr utils::TypeInfo coordination::EnableWritingOnMainReq::kType{utils::TypeId::COORD_ENABLE_WRITING_ON_MAIN_REQ,
"CoordEnableWritingOnMainReq", nullptr};
constexpr utils::TypeInfo coordination::EnableWritingOnMainRes::kType{utils::TypeId::COORD_ENABLE_WRITING_ON_MAIN_RES,
"CoordEnableWritingOnMainRes", nullptr};
constexpr utils::TypeInfo coordination::GetInstanceUUIDReq::kType{utils::TypeId::COORD_GET_UUID_REQ, "CoordGetUUIDReq",
nullptr};
constexpr utils::TypeInfo coordination::GetInstanceUUIDRes::kType{utils::TypeId::COORD_GET_UUID_RES, "CoordGetUUIDRes",
nullptr};
namespace slk {
// PromoteReplicaToMainRpc
void Save(const memgraph::coordination::PromoteReplicaToMainRes &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.success, builder);
}
@@ -143,16 +77,13 @@ void Load(memgraph::coordination::PromoteReplicaToMainRes *self, memgraph::slk::
}
void Save(const memgraph::coordination::PromoteReplicaToMainReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.main_uuid_, builder);
memgraph::slk::Save(self.replication_clients_info, builder);
}
void Load(memgraph::coordination::PromoteReplicaToMainReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->main_uuid_, reader);
memgraph::slk::Load(&self->replication_clients_info, reader);
}
// DemoteMainToReplicaRpc
void Save(const memgraph::coordination::DemoteMainToReplicaReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.replication_client_info, builder);
}
@@ -169,50 +100,6 @@ void Load(memgraph::coordination::DemoteMainToReplicaRes *self, memgraph::slk::R
memgraph::slk::Load(&self->success, reader);
}
// UnregisterReplicaRpc
void Save(memgraph::coordination::UnregisterReplicaReq const &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.instance_name, builder);
}
void Load(memgraph::coordination::UnregisterReplicaReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->instance_name, reader);
}
void Save(memgraph::coordination::UnregisterReplicaRes const &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.success, builder);
}
void Load(memgraph::coordination::UnregisterReplicaRes *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->success, reader);
}
void Save(memgraph::coordination::EnableWritingOnMainRes const &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.success, builder);
}
void Load(memgraph::coordination::EnableWritingOnMainRes *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->success, reader);
}
// GetInstanceUUIDRpc
void Save(const memgraph::coordination::GetInstanceUUIDReq & /*self*/, memgraph::slk::Builder * /*builder*/) {
/* nothing to serialize*/
}
void Load(memgraph::coordination::GetInstanceUUIDReq * /*self*/, memgraph::slk::Reader * /*reader*/) {
/* nothing to serialize*/
}
void Save(const memgraph::coordination::GetInstanceUUIDRes &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.uuid, builder);
}
void Load(memgraph::coordination::GetInstanceUUIDRes *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->uuid, reader);
}
} // namespace slk
} // namespace memgraph

View File

@@ -12,7 +12,7 @@
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_server.hpp"
#include "replication_coordination_glue/handler.hpp"
#include "replication_coordination_glue/messages.hpp"
namespace memgraph::coordination {

View File

@@ -25,7 +25,7 @@
namespace memgraph::coordination {
CoordinatorState::CoordinatorState() {
MG_ASSERT(!(FLAGS_raft_server_id && FLAGS_coordinator_server_port),
MG_ASSERT(!(FLAGS_coordinator && FLAGS_coordinator_server_port),
"Instance cannot be a coordinator and have registered coordinator server.");
spdlog::info("Executing coordinator constructor");
@@ -41,53 +41,37 @@ CoordinatorState::CoordinatorState() {
}
}
auto CoordinatorState::RegisterReplicationInstance(CoordinatorClientConfig config)
-> RegisterInstanceCoordinatorStatus {
MG_ASSERT(std::holds_alternative<CoordinatorInstance>(data_),
auto CoordinatorState::RegisterInstance(CoordinatorClientConfig config) -> RegisterInstanceCoordinatorStatus {
MG_ASSERT(std::holds_alternative<CoordinatorData>(data_),
"Coordinator cannot register replica since variant holds wrong alternative");
return std::visit(
memgraph::utils::Overloaded{[](const CoordinatorMainReplicaData & /*coordinator_main_replica_data*/) {
return RegisterInstanceCoordinatorStatus::NOT_COORDINATOR;
},
[config](CoordinatorInstance &coordinator_instance) {
return coordinator_instance.RegisterReplicationInstance(config);
}},
memgraph::utils::Overloaded{
[](const CoordinatorMainReplicaData & /*coordinator_main_replica_data*/) {
return RegisterInstanceCoordinatorStatus::NOT_COORDINATOR;
},
[config](CoordinatorData &coordinator_data) { return coordinator_data.RegisterInstance(config); }},
data_);
}
auto CoordinatorState::UnregisterReplicationInstance(std::string instance_name) -> UnregisterInstanceCoordinatorStatus {
MG_ASSERT(std::holds_alternative<CoordinatorInstance>(data_),
"Coordinator cannot unregister instance since variant holds wrong alternative");
return std::visit(
memgraph::utils::Overloaded{[](const CoordinatorMainReplicaData & /*coordinator_main_replica_data*/) {
return UnregisterInstanceCoordinatorStatus::NOT_COORDINATOR;
},
[&instance_name](CoordinatorInstance &coordinator_instance) {
return coordinator_instance.UnregisterReplicationInstance(instance_name);
}},
data_);
}
auto CoordinatorState::SetReplicationInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus {
MG_ASSERT(std::holds_alternative<CoordinatorInstance>(data_),
auto CoordinatorState::SetInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus {
MG_ASSERT(std::holds_alternative<CoordinatorData>(data_),
"Coordinator cannot register replica since variant holds wrong alternative");
return std::visit(
memgraph::utils::Overloaded{[](const CoordinatorMainReplicaData & /*coordinator_main_replica_data*/) {
return SetInstanceToMainCoordinatorStatus::NOT_COORDINATOR;
},
[&instance_name](CoordinatorInstance &coordinator_instance) {
return coordinator_instance.SetReplicationInstanceToMain(instance_name);
[&instance_name](CoordinatorData &coordinator_data) {
return coordinator_data.SetInstanceToMain(instance_name);
}},
data_);
}
auto CoordinatorState::ShowInstances() const -> std::vector<InstanceStatus> {
MG_ASSERT(std::holds_alternative<CoordinatorInstance>(data_),
auto CoordinatorState::ShowInstances() const -> std::vector<CoordinatorInstanceStatus> {
MG_ASSERT(std::holds_alternative<CoordinatorData>(data_),
"Can't call show instances on data_, as variant holds wrong alternative");
return std::get<CoordinatorInstance>(data_).ShowInstances();
return std::get<CoordinatorData>(data_).ShowInstances();
}
auto CoordinatorState::GetCoordinatorServer() const -> CoordinatorServer & {
@@ -95,13 +79,5 @@ auto CoordinatorState::GetCoordinatorServer() const -> CoordinatorServer & {
"Cannot get coordinator server since variant holds wrong alternative");
return *std::get<CoordinatorMainReplicaData>(data_).coordinator_server_;
}
auto CoordinatorState::AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address)
-> void {
MG_ASSERT(std::holds_alternative<CoordinatorInstance>(data_),
"Coordinator cannot register replica since variant holds wrong alternative");
return std::get<CoordinatorInstance>(data_).AddCoordinatorInstance(raft_server_id, raft_port, raft_address);
}
} // namespace memgraph::coordination
#endif

View File

@@ -1,111 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#ifdef MG_ENTERPRISE
#include "nuraft/coordinator_state_machine.hpp"
namespace memgraph::coordination {
auto CoordinatorStateMachine::EncodeRegisterReplicationInstance(const std::string &name) -> ptr<buffer> {
std::string str_log = name + "_replica";
ptr<buffer> log = buffer::alloc(sizeof(uint32_t) + str_log.size());
buffer_serializer bs(log);
bs.put_str(str_log);
return log;
}
auto CoordinatorStateMachine::DecodeRegisterReplicationInstance(buffer &data) -> std::string {
buffer_serializer bs(data);
return bs.get_str();
}
auto CoordinatorStateMachine::pre_commit(ulong const log_idx, buffer &data) -> ptr<buffer> {
buffer_serializer bs(data);
std::string str = bs.get_str();
spdlog::info("pre_commit {} : {}", log_idx, str);
return nullptr;
}
auto CoordinatorStateMachine::commit(ulong const log_idx, buffer &data) -> ptr<buffer> {
buffer_serializer bs(data);
std::string str = bs.get_str();
spdlog::info("commit {} : {}", log_idx, str);
last_committed_idx_ = log_idx;
return nullptr;
}
auto CoordinatorStateMachine::commit_config(ulong const log_idx, ptr<cluster_config> & /*new_conf*/) -> void {
last_committed_idx_ = log_idx;
}
auto CoordinatorStateMachine::rollback(ulong const log_idx, buffer &data) -> void {
buffer_serializer bs(data);
std::string str = bs.get_str();
spdlog::info("rollback {} : {}", log_idx, str);
}
auto CoordinatorStateMachine::read_logical_snp_obj(snapshot & /*snapshot*/, void *& /*user_snp_ctx*/, ulong /*obj_id*/,
ptr<buffer> &data_out, bool &is_last_obj) -> int {
// Put dummy data.
data_out = buffer::alloc(sizeof(int32));
buffer_serializer bs(data_out);
bs.put_i32(0);
is_last_obj = true;
return 0;
}
auto CoordinatorStateMachine::save_logical_snp_obj(snapshot &s, ulong &obj_id, buffer & /*data*/, bool /*is_first_obj*/,
bool /*is_last_obj*/) -> void {
spdlog::info("save snapshot {} term {} object ID", s.get_last_log_idx(), s.get_last_log_term(), obj_id);
// Request next object.
obj_id++;
}
auto CoordinatorStateMachine::apply_snapshot(snapshot &s) -> bool {
spdlog::info("apply snapshot {} term {}", s.get_last_log_idx(), s.get_last_log_term());
{
auto lock = std::lock_guard{last_snapshot_lock_};
ptr<buffer> snp_buf = s.serialize();
last_snapshot_ = snapshot::deserialize(*snp_buf);
}
return true;
}
auto CoordinatorStateMachine::free_user_snp_ctx(void *&user_snp_ctx) -> void {}
auto CoordinatorStateMachine::last_snapshot() -> ptr<snapshot> {
auto lock = std::lock_guard{last_snapshot_lock_};
return last_snapshot_;
}
auto CoordinatorStateMachine::last_commit_index() -> ulong { return last_committed_idx_; }
auto CoordinatorStateMachine::create_snapshot(snapshot &s, async_result<bool>::handler_type &when_done) -> void {
spdlog::info("create snapshot {} term {}", s.get_last_log_idx(), s.get_last_log_term());
// Clone snapshot from `s`.
{
auto lock = std::lock_guard{last_snapshot_lock_};
ptr<buffer> snp_buf = s.serialize();
last_snapshot_ = snapshot::deserialize(*snp_buf);
}
ptr<std::exception> except(nullptr);
bool ret = true;
when_done(ret, except);
}
} // namespace memgraph::coordination
#endif

View File

@@ -1,68 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#ifdef MG_ENTERPRISE
#include "nuraft/coordinator_state_manager.hpp"
namespace memgraph::coordination {
using nuraft::cluster_config;
using nuraft::cs_new;
using nuraft::srv_config;
using nuraft::srv_state;
using nuraft::state_mgr;
CoordinatorStateManager::CoordinatorStateManager(int srv_id, std::string const &endpoint)
: my_id_(srv_id), my_endpoint_(endpoint), cur_log_store_(cs_new<CoordinatorLogStore>()) {
my_srv_config_ = cs_new<srv_config>(srv_id, endpoint);
// Initial cluster config: contains only one server (myself).
cluster_config_ = cs_new<cluster_config>();
cluster_config_->get_servers().push_back(my_srv_config_);
}
auto CoordinatorStateManager::load_config() -> ptr<cluster_config> {
// Just return in-memory data in this example.
// May require reading from disk here, if it has been written to disk.
return cluster_config_;
}
auto CoordinatorStateManager::save_config(cluster_config const &config) -> void {
// Just keep in memory in this example.
// Need to write to disk here, if want to make it durable.
ptr<buffer> buf = config.serialize();
cluster_config_ = cluster_config::deserialize(*buf);
}
auto CoordinatorStateManager::save_state(srv_state const &state) -> void {
// Just keep in memory in this example.
// Need to write to disk here, if want to make it durable.
ptr<buffer> buf = state.serialize();
saved_state_ = srv_state::deserialize(*buf);
}
auto CoordinatorStateManager::read_state() -> ptr<srv_state> {
// Just return in-memory data in this example.
// May require reading from disk here, if it has been written to disk.
return saved_state_;
}
auto CoordinatorStateManager::load_log_store() -> ptr<log_store> { return cur_log_store_; }
auto CoordinatorStateManager::server_id() -> int32 { return my_id_; }
auto CoordinatorStateManager::system_exit(int const exit_code) -> void {}
auto CoordinatorStateManager::GetSrvConfig() const -> ptr<srv_config> { return my_srv_config_; }
} // namespace memgraph::coordination
#endif

View File

@@ -1,60 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#if FMT_VERSION > 90000
#include <fmt/ostream.h>
#include <string>
#include <libnuraft/nuraft.hxx>
#include "utils/logging.hpp"
inline std::string ToString(const nuraft::cmd_result_code &code) {
switch (code) {
case nuraft::cmd_result_code::OK:
return "OK";
case nuraft::cmd_result_code::FAILED:
return "FAILED";
case nuraft::cmd_result_code::RESULT_NOT_EXIST_YET:
return "RESULT_NOT_EXIST_YET";
case nuraft::cmd_result_code::TERM_MISMATCH:
return "TERM_MISMATCH";
case nuraft::cmd_result_code::SERVER_IS_LEAVING:
return "SERVER_IS_LEAVING";
case nuraft::cmd_result_code::CANNOT_REMOVE_LEADER:
return "CANNOT_REMOVE_LEADER";
case nuraft::cmd_result_code::SERVER_NOT_FOUND:
return "SERVER_NOT_FOUND";
case nuraft::cmd_result_code::SERVER_IS_JOINING:
return "SERVER_IS_JOINING";
case nuraft::cmd_result_code::CONFIG_CHANGING:
return "CONFIG_CHANGING";
case nuraft::cmd_result_code::SERVER_ALREADY_EXISTS:
return "SERVER_ALREADY_EXISTS";
case nuraft::cmd_result_code::BAD_REQUEST:
return "BAD_REQUEST";
case nuraft::cmd_result_code::NOT_LEADER:
return "NOT_LEADER";
case nuraft::cmd_result_code::TIMEOUT:
return "TIMEOUT";
case nuraft::cmd_result_code::CANCELLED:
return "CANCELLED";
}
LOG_FATAL("ToString of a nuraft::cmd_result_code -> check missing switch case");
}
inline std::ostream &operator<<(std::ostream &os, const nuraft::cmd_result_code &code) {
os << ToString(code);
return os;
}
template <>
class fmt::formatter<nuraft::cmd_result_code> : public fmt::ostream_formatter {};
#endif

View File

@@ -9,6 +9,14 @@
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
namespace memgraph::coordination {
enum class GetInstanceUUIDError { NO_RESPONSE, RPC_EXCEPTION };
#ifdef MG_EXPERIMENTAL_HIGH_AVAILABILITY
constexpr bool allow_ha = true;
#else
constexpr bool allow_ha = false;
#endif
} // namespace memgraph::coordination

View File

@@ -15,21 +15,18 @@
#include "coordination/coordinator_config.hpp"
#include "rpc/client.hpp"
#include "rpc_errors.hpp"
#include "utils/result.hpp"
#include "utils/scheduler.hpp"
#include "utils/uuid.hpp"
namespace memgraph::coordination {
class CoordinatorInstance;
using HealthCheckCallback = std::function<void(CoordinatorInstance *, std::string_view)>;
class CoordinatorData;
using HealthCheckCallback = std::function<void(CoordinatorData *, std::string_view)>;
using ReplicationClientsInfo = std::vector<ReplClientInfo>;
class CoordinatorClient {
public:
explicit CoordinatorClient(CoordinatorInstance *coord_instance, CoordinatorClientConfig config,
HealthCheckCallback succ_cb, HealthCheckCallback fail_cb);
explicit CoordinatorClient(CoordinatorData *coord_data_, CoordinatorClientConfig config, HealthCheckCallback succ_cb,
HealthCheckCallback fail_cb);
~CoordinatorClient() = default;
@@ -47,29 +44,13 @@ class CoordinatorClient {
auto InstanceName() const -> std::string;
auto SocketAddress() const -> std::string;
[[nodiscard]] auto SendPromoteReplicaToMainRpc(ReplicationClientsInfo replication_clients_info) const -> bool;
[[nodiscard]] auto DemoteToReplica() const -> bool;
auto SendPromoteReplicaToMainRpc(const utils::UUID &uuid, ReplicationClientsInfo replication_clients_info) const
-> bool;
auto SendSwapMainUUIDRpc(const utils::UUID &uuid) const -> bool;
auto SendUnregisterReplicaRpc(std::string const &instance_name) const -> bool;
auto SendEnableWritingOnMainRpc() const -> bool;
auto SendGetInstanceUUIDRpc() const -> memgraph::utils::BasicResult<GetInstanceUUIDError, std::optional<utils::UUID>>;
auto ReplicationClientInfo() const -> ReplClientInfo;
auto SetCallbacks(HealthCheckCallback succ_cb, HealthCheckCallback fail_cb) -> void;
auto RpcClient() -> rpc::Client & { return rpc_client_; }
auto InstanceDownTimeoutSec() const -> std::chrono::seconds;
auto InstanceGetUUIDFrequencySec() const -> std::chrono::seconds;
friend bool operator==(CoordinatorClient const &first, CoordinatorClient const &second) {
return first.config_ == second.config_;
}
@@ -77,11 +58,12 @@ class CoordinatorClient {
private:
utils::Scheduler instance_checker_;
// TODO: (andi) Pimpl?
communication::ClientContext rpc_context_;
mutable rpc::Client rpc_client_;
CoordinatorClientConfig config_;
CoordinatorInstance *coord_instance_;
CoordinatorData *coord_data_;
HealthCheckCallback succ_cb_;
HealthCheckCallback fail_cb_;
};

View File

@@ -11,11 +11,12 @@
#pragma once
#if FMT_VERSION > 90000
#include <fmt/ostream.h>
#ifdef MG_ENTERPRISE
namespace memgraph::coordination {
#include "io/network/endpoint.hpp"
struct CoordinatorClusterConfig {
static constexpr int alive_response_time_difference_sec_{5};
};
template <>
class fmt::formatter<memgraph::io::network::Endpoint> : public fmt::ostream_formatter {};
} // namespace memgraph::coordination
#endif

View File

@@ -28,9 +28,7 @@ struct CoordinatorClientConfig {
std::string instance_name;
std::string ip_address;
uint16_t port{};
std::chrono::seconds instance_health_check_frequency_sec{1};
std::chrono::seconds instance_down_timeout_sec{5};
std::chrono::seconds instance_get_uuid_frequency_sec{10};
std::chrono::seconds health_check_frequency_sec{1};
auto SocketAddress() const -> std::string { return ip_address + ":" + std::to_string(port); }

View File

@@ -0,0 +1,52 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_instance.hpp"
#include "coordination/coordinator_instance_status.hpp"
#include "coordination/coordinator_server.hpp"
#include "coordination/register_main_replica_coordinator_status.hpp"
#include "utils/rw_lock.hpp"
#include "utils/thread_pool.hpp"
#include <list>
namespace memgraph::coordination {
class CoordinatorData {
public:
CoordinatorData();
[[nodiscard]] auto RegisterInstance(CoordinatorClientConfig config) -> RegisterInstanceCoordinatorStatus;
[[nodiscard]] auto SetInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus;
auto TryFailover() -> void;
auto ShowInstances() const -> std::vector<CoordinatorInstanceStatus>;
private:
auto ClusterHasAliveMain_() const -> bool;
mutable utils::RWLock coord_data_lock_{utils::RWLock::Priority::READ};
HealthCheckCallback main_succ_cb_, main_fail_cb_, replica_succ_cb_, replica_fail_cb_;
// NOTE: Must be std::list because we rely on pointer stability
std::list<CoordinatorInstance> registered_instances_;
};
struct CoordinatorMainReplicaData {
std::unique_ptr<CoordinatorServer> coordinator_server_;
};
} // namespace memgraph::coordination
#endif

View File

@@ -28,60 +28,5 @@ class CoordinatorRegisterInstanceException final : public utils::BasicException
SPECIALIZE_GET_EXCEPTION_NAME(CoordinatorRegisterInstanceException)
};
class RaftServerStartException final : public utils::BasicException {
public:
explicit RaftServerStartException(std::string_view what) noexcept : BasicException(what) {}
template <class... Args>
explicit RaftServerStartException(fmt::format_string<Args...> fmt, Args &&...args) noexcept
: RaftServerStartException(fmt::format(fmt, std::forward<Args>(args)...)) {}
SPECIALIZE_GET_EXCEPTION_NAME(RaftServerStartException)
};
class RaftAddServerException final : public utils::BasicException {
public:
explicit RaftAddServerException(std::string_view what) noexcept : BasicException(what) {}
template <class... Args>
explicit RaftAddServerException(fmt::format_string<Args...> fmt, Args &&...args) noexcept
: RaftAddServerException(fmt::format(fmt, std::forward<Args>(args)...)) {}
SPECIALIZE_GET_EXCEPTION_NAME(RaftAddServerException)
};
class RaftBecomeLeaderException final : public utils::BasicException {
public:
explicit RaftBecomeLeaderException(std::string_view what) noexcept : BasicException(what) {}
template <class... Args>
explicit RaftBecomeLeaderException(fmt::format_string<Args...> fmt, Args &&...args) noexcept
: RaftBecomeLeaderException(fmt::format(fmt, std::forward<Args>(args)...)) {}
SPECIALIZE_GET_EXCEPTION_NAME(RaftBecomeLeaderException)
};
class RaftCouldNotFindEntryException final : public utils::BasicException {
public:
explicit RaftCouldNotFindEntryException(std::string_view what) noexcept : BasicException(what) {}
template <class... Args>
explicit RaftCouldNotFindEntryException(fmt::format_string<Args...> fmt, Args &&...args) noexcept
: RaftCouldNotFindEntryException(fmt::format(fmt, std::forward<Args>(args)...)) {}
SPECIALIZE_GET_EXCEPTION_NAME(RaftCouldNotFindEntryException)
};
class RaftCouldNotParseFlagsException final : public utils::BasicException {
public:
explicit RaftCouldNotParseFlagsException(std::string_view what) noexcept : BasicException(what) {}
template <class... Args>
explicit RaftCouldNotParseFlagsException(fmt::format_string<Args...> fmt, Args &&...args) noexcept
: RaftCouldNotParseFlagsException(fmt::format(fmt, std::forward<Args>(args)...)) {}
SPECIALIZE_GET_EXCEPTION_NAME(RaftCouldNotParseFlagsException)
};
} // namespace memgraph::coordination
#endif

View File

@@ -31,16 +31,6 @@ class CoordinatorHandlers {
slk::Builder *res_builder);
static void DemoteMainToReplicaHandler(replication::ReplicationHandler &replication_handler, slk::Reader *req_reader,
slk::Builder *res_builder);
static void SwapMainUUIDHandler(replication::ReplicationHandler &replication_handler, slk::Reader *req_reader,
slk::Builder *res_builder);
static void UnregisterReplicaHandler(replication::ReplicationHandler &replication_handler, slk::Reader *req_reader,
slk::Builder *res_builder);
static void EnableWritingOnMainHandler(replication::ReplicationHandler &replication_handler, slk::Reader *req_reader,
slk::Builder *res_builder);
static void GetInstanceUUIDHandler(replication::ReplicationHandler &replication_handler, slk::Reader *req_reader,
slk::Builder *res_builder);
};
} // namespace memgraph::dbms

View File

@@ -13,47 +13,55 @@
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_server.hpp"
#include "coordination/instance_status.hpp"
#include "coordination/raft_state.hpp"
#include "coordination/register_main_replica_coordinator_status.hpp"
#include "coordination/replication_instance.hpp"
#include "utils/rw_lock.hpp"
#include "utils/thread_pool.hpp"
#include <list>
#include "coordination/coordinator_client.hpp"
#include "coordination/coordinator_cluster_config.hpp"
#include "coordination/coordinator_exceptions.hpp"
#include "replication_coordination_glue/role.hpp"
namespace memgraph::coordination {
class CoordinatorData;
class CoordinatorInstance {
public:
CoordinatorInstance();
CoordinatorInstance(CoordinatorData *data, CoordinatorClientConfig config, HealthCheckCallback succ_cb,
HealthCheckCallback fail_cb);
[[nodiscard]] auto RegisterReplicationInstance(CoordinatorClientConfig config) -> RegisterInstanceCoordinatorStatus;
[[nodiscard]] auto UnregisterReplicationInstance(std::string instance_name) -> UnregisterInstanceCoordinatorStatus;
CoordinatorInstance(CoordinatorInstance const &other) = delete;
CoordinatorInstance &operator=(CoordinatorInstance const &other) = delete;
CoordinatorInstance(CoordinatorInstance &&other) noexcept = delete;
CoordinatorInstance &operator=(CoordinatorInstance &&other) noexcept = delete;
~CoordinatorInstance() = default;
[[nodiscard]] auto SetReplicationInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus;
auto OnSuccessPing() -> void;
auto OnFailPing() -> bool;
auto ShowInstances() const -> std::vector<InstanceStatus>;
auto IsAlive() const -> bool;
auto TryFailover() -> void;
auto InstanceName() const -> std::string;
auto SocketAddress() const -> std::string;
auto AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address) -> void;
auto IsReplica() const -> bool;
auto IsMain() const -> bool;
auto GetMainUUID() const -> utils::UUID;
auto PromoteToMain(ReplicationClientsInfo repl_clients_info, HealthCheckCallback main_succ_cb,
HealthCheckCallback main_fail_cb) -> bool;
auto DemoteToReplica(HealthCheckCallback replica_succ_cb, HealthCheckCallback replica_fail_cb) -> bool;
auto SetMainUUID(utils::UUID new_uuid) -> void;
auto PauseFrequentCheck() -> void;
auto ResumeFrequentCheck() -> void;
auto ReplicationClientInfo() const -> ReplClientInfo;
private:
HealthCheckCallback main_succ_cb_, main_fail_cb_, replica_succ_cb_, replica_fail_cb_;
CoordinatorClient client_;
replication_coordination_glue::ReplicationRole replication_role_;
std::chrono::system_clock::time_point last_response_time_{};
bool is_alive_{false};
// NOTE: Must be std::list because we rely on pointer stability
std::list<ReplicationInstance> repl_instances_;
mutable utils::RWLock coord_instance_lock_{utils::RWLock::Priority::READ};
utils::UUID main_uuid_;
RaftState raft_state_;
friend bool operator==(CoordinatorInstance const &first, CoordinatorInstance const &second) {
return first.client_ == second.client_ && first.replication_role_ == second.replication_role_;
}
};
} // namespace memgraph::coordination

View File

@@ -19,13 +19,10 @@
namespace memgraph::coordination {
// TODO: (andi) For phase IV. Some instances won't have raft_socket_address, coord_socket_address, replication_role and
// cluster role... At the end, all instances will have everything.
struct InstanceStatus {
struct CoordinatorInstanceStatus {
std::string instance_name;
std::string raft_socket_address;
std::string coord_socket_address;
std::string cluster_role;
std::string socket_address;
std::string replication_role;
bool is_alive;
};

View File

@@ -11,7 +11,6 @@
#pragma once
#include "utils/uuid.hpp"
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_config.hpp"
@@ -27,13 +26,10 @@ struct PromoteReplicaToMainReq {
static void Load(PromoteReplicaToMainReq *self, memgraph::slk::Reader *reader);
static void Save(const PromoteReplicaToMainReq &self, memgraph::slk::Builder *builder);
explicit PromoteReplicaToMainReq(const utils::UUID &uuid,
std::vector<CoordinatorClientConfig::ReplicationClientInfo> replication_clients_info)
: main_uuid_(uuid), replication_clients_info(std::move(replication_clients_info)) {}
explicit PromoteReplicaToMainReq(std::vector<CoordinatorClientConfig::ReplicationClientInfo> replication_clients_info)
: replication_clients_info(std::move(replication_clients_info)) {}
PromoteReplicaToMainReq() = default;
// get uuid here
utils::UUID main_uuid_;
std::vector<CoordinatorClientConfig::ReplicationClientInfo> replication_clients_info;
};
@@ -82,116 +78,27 @@ struct DemoteMainToReplicaRes {
using DemoteMainToReplicaRpc = rpc::RequestResponse<DemoteMainToReplicaReq, DemoteMainToReplicaRes>;
struct UnregisterReplicaReq {
static const utils::TypeInfo kType;
static const utils::TypeInfo &GetTypeInfo() { return kType; }
static void Load(UnregisterReplicaReq *self, memgraph::slk::Reader *reader);
static void Save(UnregisterReplicaReq const &self, memgraph::slk::Builder *builder);
explicit UnregisterReplicaReq(std::string instance_name) : instance_name(std::move(instance_name)) {}
UnregisterReplicaReq() = default;
std::string instance_name;
};
struct UnregisterReplicaRes {
static const utils::TypeInfo kType;
static const utils::TypeInfo &GetTypeInfo() { return kType; }
static void Load(UnregisterReplicaRes *self, memgraph::slk::Reader *reader);
static void Save(const UnregisterReplicaRes &self, memgraph::slk::Builder *builder);
explicit UnregisterReplicaRes(bool success) : success(success) {}
UnregisterReplicaRes() = default;
bool success;
};
using UnregisterReplicaRpc = rpc::RequestResponse<UnregisterReplicaReq, UnregisterReplicaRes>;
struct EnableWritingOnMainReq {
static const utils::TypeInfo kType;
static const utils::TypeInfo &GetTypeInfo() { return kType; }
static void Load(EnableWritingOnMainReq *self, memgraph::slk::Reader *reader);
static void Save(EnableWritingOnMainReq const &self, memgraph::slk::Builder *builder);
EnableWritingOnMainReq() = default;
};
struct EnableWritingOnMainRes {
static const utils::TypeInfo kType;
static const utils::TypeInfo &GetTypeInfo() { return kType; }
static void Load(EnableWritingOnMainRes *self, memgraph::slk::Reader *reader);
static void Save(EnableWritingOnMainRes const &self, memgraph::slk::Builder *builder);
explicit EnableWritingOnMainRes(bool success) : success(success) {}
EnableWritingOnMainRes() = default;
bool success;
};
using EnableWritingOnMainRpc = rpc::RequestResponse<EnableWritingOnMainReq, EnableWritingOnMainRes>;
struct GetInstanceUUIDReq {
static const utils::TypeInfo kType;
static const utils::TypeInfo &GetTypeInfo() { return kType; }
static void Load(GetInstanceUUIDReq *self, memgraph::slk::Reader *reader);
static void Save(const GetInstanceUUIDReq &self, memgraph::slk::Builder *builder);
GetInstanceUUIDReq() = default;
};
struct GetInstanceUUIDRes {
static const utils::TypeInfo kType;
static const utils::TypeInfo &GetTypeInfo() { return kType; }
static void Load(GetInstanceUUIDRes *self, memgraph::slk::Reader *reader);
static void Save(const GetInstanceUUIDRes &self, memgraph::slk::Builder *builder);
explicit GetInstanceUUIDRes(std::optional<utils::UUID> uuid) : uuid(uuid) {}
GetInstanceUUIDRes() = default;
std::optional<utils::UUID> uuid;
};
using GetInstanceUUIDRpc = rpc::RequestResponse<GetInstanceUUIDReq, GetInstanceUUIDRes>;
} // namespace memgraph::coordination
// SLK serialization declarations
namespace memgraph::slk {
// PromoteReplicaToMainRpc
void Save(const memgraph::coordination::PromoteReplicaToMainRes &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::PromoteReplicaToMainRes *self, memgraph::slk::Reader *reader);
void Save(const memgraph::coordination::PromoteReplicaToMainReq &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::PromoteReplicaToMainReq *self, memgraph::slk::Reader *reader);
// DemoteMainToReplicaRpc
void Save(const memgraph::coordination::DemoteMainToReplicaRes &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::DemoteMainToReplicaRes *self, memgraph::slk::Reader *reader);
void Save(const memgraph::coordination::DemoteMainToReplicaReq &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::DemoteMainToReplicaReq *self, memgraph::slk::Reader *reader);
// GetInstanceUUIDRpc
void Save(const memgraph::coordination::GetInstanceUUIDReq &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::GetInstanceUUIDReq *self, memgraph::slk::Reader *reader);
void Save(const memgraph::coordination::GetInstanceUUIDRes &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::GetInstanceUUIDRes *self, memgraph::slk::Reader *reader);
// UnregisterReplicaRpc
void Save(memgraph::coordination::UnregisterReplicaRes const &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::UnregisterReplicaRes *self, memgraph::slk::Reader *reader);
void Save(memgraph::coordination::UnregisterReplicaReq const &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::UnregisterReplicaReq *self, memgraph::slk::Reader *reader);
void Save(memgraph::coordination::EnableWritingOnMainRes const &self, memgraph::slk::Builder *builder);
void Load(memgraph::coordination::EnableWritingOnMainRes *self, memgraph::slk::Reader *reader);
} // namespace memgraph::slk
#endif

View File

@@ -13,9 +13,9 @@
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_instance.hpp"
#include "coordination/coordinator_data.hpp"
#include "coordination/coordinator_instance_status.hpp"
#include "coordination/coordinator_server.hpp"
#include "coordination/instance_status.hpp"
#include "coordination/register_main_replica_coordinator_status.hpp"
#include <variant>
@@ -33,24 +33,17 @@ class CoordinatorState {
CoordinatorState(CoordinatorState &&) noexcept = delete;
CoordinatorState &operator=(CoordinatorState &&) noexcept = delete;
[[nodiscard]] auto RegisterReplicationInstance(CoordinatorClientConfig config) -> RegisterInstanceCoordinatorStatus;
[[nodiscard]] auto UnregisterReplicationInstance(std::string instance_name) -> UnregisterInstanceCoordinatorStatus;
[[nodiscard]] auto RegisterInstance(CoordinatorClientConfig config) -> RegisterInstanceCoordinatorStatus;
[[nodiscard]] auto SetReplicationInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus;
[[nodiscard]] auto SetInstanceToMain(std::string instance_name) -> SetInstanceToMainCoordinatorStatus;
auto ShowInstances() const -> std::vector<InstanceStatus>;
auto ShowInstances() const -> std::vector<CoordinatorInstanceStatus>;
auto AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address) -> void;
// NOTE: The client code must check that the server exists before calling this method.
// The client code must check that the server exists before calling this method.
auto GetCoordinatorServer() const -> CoordinatorServer &;
private:
struct CoordinatorMainReplicaData {
std::unique_ptr<CoordinatorServer> coordinator_server_;
};
std::variant<CoordinatorInstance, CoordinatorMainReplicaData> data_;
std::variant<CoordinatorData, CoordinatorMainReplicaData> data_;
};
} // namespace memgraph::coordination

View File

@@ -1,79 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#ifdef MG_ENTERPRISE
#include <flags/replication.hpp>
#include <libnuraft/nuraft.hxx>
namespace memgraph::coordination {
using BecomeLeaderCb = std::function<void()>;
using BecomeFollowerCb = std::function<void()>;
using nuraft::buffer;
using nuraft::logger;
using nuraft::ptr;
using nuraft::raft_launcher;
using nuraft::raft_server;
using nuraft::srv_config;
using nuraft::state_machine;
using nuraft::state_mgr;
using raft_result = nuraft::cmd_result<ptr<buffer>>;
class RaftState {
private:
explicit RaftState(BecomeLeaderCb become_leader_cb, BecomeFollowerCb become_follower_cb, uint32_t raft_server_id,
uint32_t raft_port, std::string raft_address);
auto InitRaftServer() -> void;
public:
RaftState() = delete;
RaftState(RaftState const &other) = default;
RaftState &operator=(RaftState const &other) = default;
RaftState(RaftState &&other) noexcept = default;
RaftState &operator=(RaftState &&other) noexcept = default;
~RaftState();
static auto MakeRaftState(BecomeLeaderCb become_leader_cb, BecomeFollowerCb become_follower_cb) -> RaftState;
auto InstanceName() const -> std::string;
auto RaftSocketAddress() const -> std::string;
auto AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address) -> void;
auto GetAllCoordinators() const -> std::vector<ptr<srv_config>>;
auto RequestLeadership() -> bool;
auto IsLeader() const -> bool;
auto AppendRegisterReplicationInstance(std::string const &instance) -> ptr<raft_result>;
// TODO: (andi) I think variables below can be abstracted
uint32_t raft_server_id_;
uint32_t raft_port_;
std::string raft_address_;
ptr<state_machine> state_machine_;
ptr<state_mgr> state_manager_;
ptr<raft_server> raft_server_;
ptr<logger> logger_;
raft_launcher launcher_;
BecomeLeaderCb become_leader_cb_;
BecomeFollowerCb become_follower_cb_;
};
} // namespace memgraph::coordination
#endif

View File

@@ -22,28 +22,14 @@ enum class RegisterInstanceCoordinatorStatus : uint8_t {
ENDPOINT_EXISTS,
NOT_COORDINATOR,
RPC_FAILED,
NOT_LEADER,
RAFT_COULD_NOT_ACCEPT,
RAFT_COULD_NOT_APPEND,
SUCCESS
};
enum class UnregisterInstanceCoordinatorStatus : uint8_t {
NO_INSTANCE_WITH_NAME,
IS_MAIN,
NOT_COORDINATOR,
NOT_LEADER,
RPC_FAILED,
SUCCESS,
};
enum class SetInstanceToMainCoordinatorStatus : uint8_t {
NO_INSTANCE_WITH_NAME,
MAIN_ALREADY_EXISTS,
NOT_COORDINATOR,
SUCCESS,
COULD_NOT_PROMOTE_TO_MAIN,
SWAP_UUID_FAILED
};
} // namespace memgraph::coordination

View File

@@ -1,99 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_client.hpp"
#include "coordination/coordinator_exceptions.hpp"
#include "replication_coordination_glue/role.hpp"
#include <libnuraft/nuraft.hxx>
#include "utils/result.hpp"
#include "utils/uuid.hpp"
namespace memgraph::coordination {
class CoordinatorInstance;
class ReplicationInstance {
public:
ReplicationInstance(CoordinatorInstance *peer, CoordinatorClientConfig config, HealthCheckCallback succ_cb,
HealthCheckCallback fail_cb);
ReplicationInstance(ReplicationInstance const &other) = delete;
ReplicationInstance &operator=(ReplicationInstance const &other) = delete;
ReplicationInstance(ReplicationInstance &&other) noexcept = delete;
ReplicationInstance &operator=(ReplicationInstance &&other) noexcept = delete;
~ReplicationInstance() = default;
auto OnSuccessPing() -> void;
auto OnFailPing() -> bool;
auto IsReadyForUUIDPing() -> bool;
void UpdateReplicaLastResponseUUID();
auto IsAlive() const -> bool;
auto InstanceName() const -> std::string;
auto SocketAddress() const -> std::string;
auto IsReplica() const -> bool;
auto IsMain() const -> bool;
auto PromoteToMain(utils::UUID uuid, ReplicationClientsInfo repl_clients_info, HealthCheckCallback main_succ_cb,
HealthCheckCallback main_fail_cb) -> bool;
auto DemoteToReplica(HealthCheckCallback replica_succ_cb, HealthCheckCallback replica_fail_cb) -> bool;
auto StartFrequentCheck() -> void;
auto StopFrequentCheck() -> void;
auto PauseFrequentCheck() -> void;
auto ResumeFrequentCheck() -> void;
auto ReplicationClientInfo() const -> ReplClientInfo;
auto EnsureReplicaHasCorrectMainUUID(utils::UUID const &curr_main_uuid) -> bool;
auto SendSwapAndUpdateUUID(const utils::UUID &new_main_uuid) -> bool;
auto SendUnregisterReplicaRpc(std::string const &instance_name) -> bool;
auto SendGetInstanceUUID() -> utils::BasicResult<coordination::GetInstanceUUIDError, std::optional<utils::UUID>>;
auto GetClient() -> CoordinatorClient &;
auto EnableWritingOnMain() -> bool;
auto SetNewMainUUID(utils::UUID const &main_uuid) -> void;
auto ResetMainUUID() -> void;
auto GetMainUUID() const -> const std::optional<utils::UUID> &;
private:
CoordinatorClient client_;
replication_coordination_glue::ReplicationRole replication_role_;
std::chrono::system_clock::time_point last_response_time_{};
bool is_alive_{false};
std::chrono::system_clock::time_point last_check_of_uuid_{};
// for replica this is main uuid of current main
// for "main" main this same as in CoordinatorData
// it is set to nullopt when replica is down
// TLDR; when replica is down and comes back up we reset uuid of main replica is listening to
// so we need to send swap uuid again
std::optional<utils::UUID> main_uuid_;
friend bool operator==(ReplicationInstance const &first, ReplicationInstance const &second) {
return first.client_ == second.client_ && first.replication_role_ == second.replication_role_;
}
};
} // namespace memgraph::coordination
#endif

View File

@@ -1,70 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#ifdef MG_ENTERPRISE
#include <libnuraft/nuraft.hxx>
namespace memgraph::coordination {
using nuraft::buffer;
using nuraft::int32;
using nuraft::int64;
using nuraft::log_entry;
using nuraft::log_store;
using nuraft::ptr;
using nuraft::raft_server;
class CoordinatorLogStore : public log_store {
public:
CoordinatorLogStore();
CoordinatorLogStore(CoordinatorLogStore const &) = delete;
CoordinatorLogStore &operator=(CoordinatorLogStore const &) = delete;
CoordinatorLogStore(CoordinatorLogStore &&) = delete;
CoordinatorLogStore &operator=(CoordinatorLogStore &&) = delete;
~CoordinatorLogStore() override;
ulong next_slot() const override;
ulong start_index() const override;
ptr<log_entry> last_entry() const override;
ulong append(ptr<log_entry> &entry) override;
void write_at(ulong index, ptr<log_entry> &entry) override;
ptr<std::vector<ptr<log_entry>>> log_entries(ulong start, ulong end) override;
ptr<log_entry> entry_at(ulong index) override;
ulong term_at(ulong index) override;
ptr<buffer> pack(ulong index, int32 cnt) override;
void apply_pack(ulong index, buffer &pack) override;
bool compact(ulong last_log_index) override;
bool flush() override;
private:
auto FindOrDefault_(ulong index) const -> ptr<log_entry>;
std::map<ulong, ptr<log_entry>> logs_;
mutable std::mutex logs_lock_;
std::atomic<ulong> start_idx_;
};
} // namespace memgraph::coordination
#endif

View File

@@ -1,76 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#ifdef MG_ENTERPRISE
#include <spdlog/spdlog.h>
#include <libnuraft/nuraft.hxx>
namespace memgraph::coordination {
using nuraft::async_result;
using nuraft::buffer;
using nuraft::buffer_serializer;
using nuraft::cluster_config;
using nuraft::int32;
using nuraft::ptr;
using nuraft::snapshot;
using nuraft::state_machine;
class CoordinatorStateMachine : public state_machine {
public:
CoordinatorStateMachine() = default;
CoordinatorStateMachine(CoordinatorStateMachine const &) = delete;
CoordinatorStateMachine &operator=(CoordinatorStateMachine const &) = delete;
CoordinatorStateMachine(CoordinatorStateMachine &&) = delete;
CoordinatorStateMachine &operator=(CoordinatorStateMachine &&) = delete;
~CoordinatorStateMachine() override {}
static auto EncodeRegisterReplicationInstance(const std::string &name) -> ptr<buffer>;
static auto DecodeRegisterReplicationInstance(buffer &data) -> std::string;
auto pre_commit(ulong log_idx, buffer &data) -> ptr<buffer> override;
auto commit(ulong log_idx, buffer &data) -> ptr<buffer> override;
auto commit_config(ulong log_idx, ptr<cluster_config> & /*new_conf*/) -> void override;
auto rollback(ulong log_idx, buffer &data) -> void override;
auto read_logical_snp_obj(snapshot & /*snapshot*/, void *& /*user_snp_ctx*/, ulong /*obj_id*/, ptr<buffer> &data_out,
bool &is_last_obj) -> int override;
auto save_logical_snp_obj(snapshot &s, ulong &obj_id, buffer & /*data*/, bool /*is_first_obj*/, bool /*is_last_obj*/)
-> void override;
auto apply_snapshot(snapshot &s) -> bool override;
auto free_user_snp_ctx(void *&user_snp_ctx) -> void override;
auto last_snapshot() -> ptr<snapshot> override;
auto last_commit_index() -> ulong override;
auto create_snapshot(snapshot &s, async_result<bool>::handler_type &when_done) -> void override;
private:
std::atomic<uint64_t> last_committed_idx_{0};
ptr<snapshot> last_snapshot_;
std::mutex last_snapshot_lock_;
};
} // namespace memgraph::coordination
#endif

View File

@@ -1,66 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#ifdef MG_ENTERPRISE
#include "nuraft/coordinator_log_store.hpp"
#include <spdlog/spdlog.h>
#include <libnuraft/nuraft.hxx>
namespace memgraph::coordination {
using nuraft::cluster_config;
using nuraft::cs_new;
using nuraft::srv_config;
using nuraft::srv_state;
using nuraft::state_mgr;
class CoordinatorStateManager : public state_mgr {
public:
explicit CoordinatorStateManager(int srv_id, std::string const &endpoint);
CoordinatorStateManager(CoordinatorStateManager const &) = delete;
CoordinatorStateManager &operator=(CoordinatorStateManager const &) = delete;
CoordinatorStateManager(CoordinatorStateManager &&) = delete;
CoordinatorStateManager &operator=(CoordinatorStateManager &&) = delete;
~CoordinatorStateManager() override = default;
auto load_config() -> ptr<cluster_config> override;
auto save_config(cluster_config const &config) -> void override;
auto save_state(srv_state const &state) -> void override;
auto read_state() -> ptr<srv_state> override;
auto load_log_store() -> ptr<log_store> override;
auto server_id() -> int32 override;
auto system_exit(int exit_code) -> void override;
auto GetSrvConfig() const -> ptr<srv_config>;
private:
int my_id_;
std::string my_endpoint_;
ptr<CoordinatorLogStore> cur_log_store_;
ptr<srv_config> my_srv_config_;
ptr<cluster_config> cluster_config_;
ptr<srv_state> saved_state_;
};
} // namespace memgraph::coordination
#endif

View File

@@ -1,140 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#ifdef MG_ENTERPRISE
#include "coordination/raft_state.hpp"
#include "coordination/coordinator_exceptions.hpp"
#include "nuraft/coordinator_state_machine.hpp"
#include "nuraft/coordinator_state_manager.hpp"
#include "utils/counter.hpp"
namespace memgraph::coordination {
using nuraft::asio_service;
using nuraft::cb_func;
using nuraft::CbReturnCode;
using nuraft::cmd_result;
using nuraft::cs_new;
using nuraft::ptr;
using nuraft::raft_params;
using nuraft::raft_server;
using nuraft::srv_config;
using raft_result = cmd_result<ptr<buffer>>;
RaftState::RaftState(BecomeLeaderCb become_leader_cb, BecomeFollowerCb become_follower_cb, uint32_t raft_server_id,
uint32_t raft_port, std::string raft_address)
: raft_server_id_(raft_server_id),
raft_port_(raft_port),
raft_address_(std::move(raft_address)),
state_machine_(cs_new<CoordinatorStateMachine>()),
state_manager_(
cs_new<CoordinatorStateManager>(raft_server_id_, raft_address_ + ":" + std::to_string(raft_port_))),
logger_(nullptr),
become_leader_cb_(std::move(become_leader_cb)),
become_follower_cb_(std::move(become_follower_cb)) {}
auto RaftState::InitRaftServer() -> void {
asio_service::options asio_opts;
asio_opts.thread_pool_size_ = 1; // TODO: (andi) Improve this
raft_params params;
params.heart_beat_interval_ = 100;
params.election_timeout_lower_bound_ = 200;
params.election_timeout_upper_bound_ = 400;
// 5 logs are preserved before the last snapshot
params.reserved_log_items_ = 5;
// Create snapshot for every 5 log appends
params.snapshot_distance_ = 5;
params.client_req_timeout_ = 3000;
params.return_method_ = raft_params::blocking;
raft_server::init_options init_opts;
init_opts.raft_callback_ = [this](cb_func::Type event_type, cb_func::Param *param) -> nuraft::CbReturnCode {
if (event_type == cb_func::BecomeLeader) {
spdlog::info("Node {} became leader", param->leaderId);
become_leader_cb_();
} else if (event_type == cb_func::BecomeFollower) {
spdlog::info("Node {} became follower", param->myId);
become_follower_cb_();
}
return CbReturnCode::Ok;
};
raft_launcher launcher;
raft_server_ = launcher.init(state_machine_, state_manager_, logger_, static_cast<int>(raft_port_), asio_opts, params,
init_opts);
if (!raft_server_) {
throw RaftServerStartException("Failed to launch raft server on {}:{}", raft_address_, raft_port_);
}
auto maybe_stop = utils::ResettableCounter<20>();
do {
if (raft_server_->is_initialized()) {
return;
}
std::this_thread::sleep_for(std::chrono::milliseconds(250));
} while (!maybe_stop());
throw RaftServerStartException("Failed to initialize raft server on {}:{}", raft_address_, raft_port_);
}
auto RaftState::MakeRaftState(BecomeLeaderCb become_leader_cb, BecomeFollowerCb become_follower_cb) -> RaftState {
uint32_t raft_server_id{0};
uint32_t raft_port{0};
try {
raft_server_id = FLAGS_raft_server_id;
raft_port = FLAGS_raft_server_port;
} catch (std::exception const &e) {
throw RaftCouldNotParseFlagsException("Failed to parse flags: {}", e.what());
}
auto raft_state =
RaftState(std::move(become_leader_cb), std::move(become_follower_cb), raft_server_id, raft_port, "127.0.0.1");
raft_state.InitRaftServer();
return raft_state;
}
RaftState::~RaftState() { launcher_.shutdown(); }
auto RaftState::InstanceName() const -> std::string { return "coordinator_" + std::to_string(raft_server_id_); }
auto RaftState::RaftSocketAddress() const -> std::string { return raft_address_ + ":" + std::to_string(raft_port_); }
auto RaftState::AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address) -> void {
auto const endpoint = raft_address + ":" + std::to_string(raft_port);
srv_config const srv_config_to_add(static_cast<int>(raft_server_id), endpoint);
if (!raft_server_->add_srv(srv_config_to_add)->get_accepted()) {
throw RaftAddServerException("Failed to add server {} to the cluster", endpoint);
}
spdlog::info("Request to add server {} to the cluster accepted", endpoint);
}
auto RaftState::GetAllCoordinators() const -> std::vector<ptr<srv_config>> {
std::vector<ptr<srv_config>> all_srv_configs;
raft_server_->get_srv_config_all(all_srv_configs);
return all_srv_configs;
}
auto RaftState::IsLeader() const -> bool { return raft_server_->is_leader(); }
auto RaftState::RequestLeadership() -> bool { return raft_server_->is_leader() || raft_server_->request_leadership(); }
auto RaftState::AppendRegisterReplicationInstance(std::string const &instance) -> ptr<raft_result> {
auto new_log = CoordinatorStateMachine::EncodeRegisterReplicationInstance(instance);
return raft_server_->append_entries({new_log});
}
} // namespace memgraph::coordination
#endif

View File

@@ -1,138 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#ifdef MG_ENTERPRISE
#include "coordination/replication_instance.hpp"
#include "replication_coordination_glue/handler.hpp"
#include "utils/result.hpp"
namespace memgraph::coordination {
ReplicationInstance::ReplicationInstance(CoordinatorInstance *peer, CoordinatorClientConfig config,
HealthCheckCallback succ_cb, HealthCheckCallback fail_cb)
: client_(peer, std::move(config), std::move(succ_cb), std::move(fail_cb)),
replication_role_(replication_coordination_glue::ReplicationRole::REPLICA) {
if (!client_.DemoteToReplica()) {
throw CoordinatorRegisterInstanceException("Failed to demote instance {} to replica", client_.InstanceName());
}
client_.StartFrequentCheck();
}
auto ReplicationInstance::OnSuccessPing() -> void {
last_response_time_ = std::chrono::system_clock::now();
is_alive_ = true;
}
auto ReplicationInstance::OnFailPing() -> bool {
auto elapsed_time = std::chrono::system_clock::now() - last_response_time_;
is_alive_ = elapsed_time < client_.InstanceDownTimeoutSec();
return is_alive_;
}
auto ReplicationInstance::IsReadyForUUIDPing() -> bool {
return std::chrono::duration_cast<std::chrono::seconds>(std::chrono::system_clock::now() - last_check_of_uuid_) >
client_.InstanceGetUUIDFrequencySec();
}
auto ReplicationInstance::InstanceName() const -> std::string { return client_.InstanceName(); }
auto ReplicationInstance::SocketAddress() const -> std::string { return client_.SocketAddress(); }
auto ReplicationInstance::IsAlive() const -> bool { return is_alive_; }
auto ReplicationInstance::IsReplica() const -> bool {
return replication_role_ == replication_coordination_glue::ReplicationRole::REPLICA;
}
auto ReplicationInstance::IsMain() const -> bool {
return replication_role_ == replication_coordination_glue::ReplicationRole::MAIN;
}
auto ReplicationInstance::PromoteToMain(utils::UUID new_uuid, ReplicationClientsInfo repl_clients_info,
HealthCheckCallback main_succ_cb, HealthCheckCallback main_fail_cb) -> bool {
if (!client_.SendPromoteReplicaToMainRpc(new_uuid, std::move(repl_clients_info))) {
return false;
}
replication_role_ = replication_coordination_glue::ReplicationRole::MAIN;
main_uuid_ = new_uuid;
client_.SetCallbacks(std::move(main_succ_cb), std::move(main_fail_cb));
return true;
}
auto ReplicationInstance::DemoteToReplica(HealthCheckCallback replica_succ_cb, HealthCheckCallback replica_fail_cb)
-> bool {
if (!client_.DemoteToReplica()) {
return false;
}
replication_role_ = replication_coordination_glue::ReplicationRole::REPLICA;
client_.SetCallbacks(std::move(replica_succ_cb), std::move(replica_fail_cb));
return true;
}
auto ReplicationInstance::StartFrequentCheck() -> void { client_.StartFrequentCheck(); }
auto ReplicationInstance::StopFrequentCheck() -> void { client_.StopFrequentCheck(); }
auto ReplicationInstance::PauseFrequentCheck() -> void { client_.PauseFrequentCheck(); }
auto ReplicationInstance::ResumeFrequentCheck() -> void { client_.ResumeFrequentCheck(); }
auto ReplicationInstance::ReplicationClientInfo() const -> CoordinatorClientConfig::ReplicationClientInfo {
return client_.ReplicationClientInfo();
}
auto ReplicationInstance::GetClient() -> CoordinatorClient & { return client_; }
auto ReplicationInstance::SetNewMainUUID(utils::UUID const &main_uuid) -> void { main_uuid_ = main_uuid; }
auto ReplicationInstance::ResetMainUUID() -> void { main_uuid_ = std::nullopt; }
auto ReplicationInstance::GetMainUUID() const -> std::optional<utils::UUID> const & { return main_uuid_; }
auto ReplicationInstance::EnsureReplicaHasCorrectMainUUID(utils::UUID const &curr_main_uuid) -> bool {
if (!IsReadyForUUIDPing()) {
return true;
}
auto res = SendGetInstanceUUID();
if (res.HasError()) {
return false;
}
UpdateReplicaLastResponseUUID();
if (res.GetValue().has_value() && res.GetValue().value() == curr_main_uuid) {
return true;
}
return SendSwapAndUpdateUUID(curr_main_uuid);
}
auto ReplicationInstance::SendSwapAndUpdateUUID(const utils::UUID &new_main_uuid) -> bool {
if (!replication_coordination_glue::SendSwapMainUUIDRpc(client_.RpcClient(), new_main_uuid)) {
return false;
}
SetNewMainUUID(new_main_uuid);
return true;
}
auto ReplicationInstance::SendUnregisterReplicaRpc(std::string const &instance_name) -> bool {
return client_.SendUnregisterReplicaRpc(instance_name);
}
auto ReplicationInstance::EnableWritingOnMain() -> bool { return client_.SendEnableWritingOnMainRpc(); }
auto ReplicationInstance::SendGetInstanceUUID()
-> utils::BasicResult<coordination::GetInstanceUUIDError, std::optional<utils::UUID>> {
return client_.SendGetInstanceUUIDRpc();
}
void ReplicationInstance::UpdateReplicaLastResponseUUID() { last_check_of_uuid_ = std::chrono::system_clock::now(); }
} // namespace memgraph::coordination
#endif

View File

@@ -16,4 +16,10 @@ namespace memgraph::dbms {
constexpr std::string_view kDefaultDB = "memgraph"; //!< Name of the default database
constexpr std::string_view kMultiTenantDir = "databases"; //!< Name of the multi-tenant directory
#ifdef MG_EXPERIMENTAL_REPLICATION_MULTITENANCY
constexpr bool allow_mt_repl = true;
#else
constexpr bool allow_mt_repl = false;
#endif
} // namespace memgraph::dbms

View File

@@ -9,41 +9,31 @@
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#include "coordination/register_main_replica_coordinator_status.hpp"
#ifdef MG_ENTERPRISE
#include "dbms/coordinator_handler.hpp"
#include "coordination/register_main_replica_coordinator_status.hpp"
#include "dbms/dbms_handler.hpp"
namespace memgraph::dbms {
CoordinatorHandler::CoordinatorHandler(coordination::CoordinatorState &coordinator_state)
: coordinator_state_(coordinator_state) {}
auto CoordinatorHandler::RegisterReplicationInstance(memgraph::coordination::CoordinatorClientConfig config)
auto CoordinatorHandler::RegisterInstance(memgraph::coordination::CoordinatorClientConfig config)
-> coordination::RegisterInstanceCoordinatorStatus {
return coordinator_state_.RegisterReplicationInstance(config);
return coordinator_state_.RegisterInstance(config);
}
auto CoordinatorHandler::UnregisterReplicationInstance(std::string instance_name)
-> coordination::UnregisterInstanceCoordinatorStatus {
return coordinator_state_.UnregisterReplicationInstance(std::move(instance_name));
}
auto CoordinatorHandler::SetReplicationInstanceToMain(std::string instance_name)
auto CoordinatorHandler::SetInstanceToMain(std::string instance_name)
-> coordination::SetInstanceToMainCoordinatorStatus {
return coordinator_state_.SetReplicationInstanceToMain(std::move(instance_name));
return coordinator_state_.SetInstanceToMain(std::move(instance_name));
}
auto CoordinatorHandler::ShowInstances() const -> std::vector<coordination::InstanceStatus> {
auto CoordinatorHandler::ShowInstances() const -> std::vector<coordination::CoordinatorInstanceStatus> {
return coordinator_state_.ShowInstances();
}
auto CoordinatorHandler::AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address)
-> void {
coordinator_state_.AddCoordinatorInstance(raft_server_id, raft_port, std::move(raft_address));
}
} // namespace memgraph::dbms
#endif

View File

@@ -14,8 +14,8 @@
#ifdef MG_ENTERPRISE
#include "coordination/coordinator_config.hpp"
#include "coordination/coordinator_instance_status.hpp"
#include "coordination/coordinator_state.hpp"
#include "coordination/instance_status.hpp"
#include "coordination/register_main_replica_coordinator_status.hpp"
#include <vector>
@@ -28,18 +28,12 @@ class CoordinatorHandler {
public:
explicit CoordinatorHandler(coordination::CoordinatorState &coordinator_state);
// TODO: (andi) When moving coordinator state on same instances, rename from RegisterReplicationInstance to
// RegisterInstance
auto RegisterReplicationInstance(coordination::CoordinatorClientConfig config)
auto RegisterInstance(coordination::CoordinatorClientConfig config)
-> coordination::RegisterInstanceCoordinatorStatus;
auto UnregisterReplicationInstance(std::string instance_name) -> coordination::UnregisterInstanceCoordinatorStatus;
auto SetInstanceToMain(std::string instance_name) -> coordination::SetInstanceToMainCoordinatorStatus;
auto SetReplicationInstanceToMain(std::string instance_name) -> coordination::SetInstanceToMainCoordinatorStatus;
auto ShowInstances() const -> std::vector<coordination::InstanceStatus>;
auto AddCoordinatorInstance(uint32_t raft_server_id, uint32_t raft_port, std::string raft_address) -> void;
auto ShowInstances() const -> std::vector<coordination::CoordinatorInstanceStatus>;
private:
coordination::CoordinatorState &coordinator_state_;

View File

@@ -16,7 +16,6 @@
#include "dbms/constants.hpp"
#include "dbms/global.hpp"
#include "flags/experimental.hpp"
#include "spdlog/spdlog.h"
#include "system/include/system/system.hpp"
#include "utils/exceptions.hpp"
@@ -39,8 +38,6 @@ std::string RegisterReplicaErrorToString(query::RegisterReplicaError error) {
return "CONNECTION_FAILED";
case COULD_NOT_BE_PERSISTED:
return "COULD_NOT_BE_PERSISTED";
case ERROR_ACCEPTING_MAIN:
return "ERROR_ACCEPTING_MAIN";
}
}
@@ -55,7 +52,7 @@ void RestoreReplication(replication::RoleMainData &mainData, DatabaseAccess db_a
spdlog::info("Replica {} restoration started for {}.", instance_client.name_, db_acc->name());
const auto &ret = db_acc->storage()->repl_storage_state_.replication_clients_.WithLock(
[&, db_acc](auto &storage_clients) mutable -> utils::BasicResult<query::RegisterReplicaError> {
auto client = std::make_unique<storage::ReplicationStorageClient>(instance_client, mainData.uuid_);
auto client = std::make_unique<storage::ReplicationStorageClient>(instance_client);
auto *storage = db_acc->storage();
client->Start(storage, std::move(db_acc));
// After start the storage <-> replica state should be READY or RECOVERING (if correctly started)
@@ -159,9 +156,9 @@ struct Durability {
}
};
DbmsHandler::DbmsHandler(storage::Config config, replication::ReplicationState &repl_state, auth::SynchedAuth &auth,
bool recovery_on_startup)
: default_config_{std::move(config)}, auth_{auth}, repl_state_{repl_state} {
DbmsHandler::DbmsHandler(storage::Config config, memgraph::system::System &system,
replication::ReplicationState &repl_state, auth::SynchedAuth &auth, bool recovery_on_startup)
: default_config_{std::move(config)}, auth_{auth}, repl_state_{repl_state}, system_{&system} {
// TODO: Decouple storage config from dbms config
// TODO: Save individual db configs inside the kvstore and restore from there
@@ -242,16 +239,14 @@ struct DropDatabase : memgraph::system::ISystemAction {
void DoDurability() override { /* Done during DBMS execution */
}
bool DoReplication(replication::ReplicationClient &client, const utils::UUID &main_uuid,
replication::ReplicationEpoch const &epoch,
bool DoReplication(replication::ReplicationClient &client, replication::ReplicationEpoch const &epoch,
memgraph::system::Transaction const &txn) const override {
auto check_response = [](const storage::replication::DropDatabaseRes &response) {
return response.result != storage::replication::DropDatabaseRes::Result::FAILURE;
};
return client.SteamAndFinalizeDelta<storage::replication::DropDatabaseRpc>(
check_response, main_uuid, std::string(epoch.id()), txn.last_committed_system_timestamp(), txn.timestamp(),
uuid_);
check_response, epoch.id(), txn.last_committed_system_timestamp(), txn.timestamp(), uuid_);
}
void PostReplication(replication::RoleMainData &mainData) const override {}
@@ -328,16 +323,14 @@ struct CreateDatabase : memgraph::system::ISystemAction {
// Done during dbms execution
}
bool DoReplication(replication::ReplicationClient &client, const utils::UUID &main_uuid,
replication::ReplicationEpoch const &epoch,
bool DoReplication(replication::ReplicationClient &client, replication::ReplicationEpoch const &epoch,
memgraph::system::Transaction const &txn) const override {
auto check_response = [](const storage::replication::CreateDatabaseRes &response) {
return response.result != storage::replication::CreateDatabaseRes::Result::FAILURE;
};
return client.SteamAndFinalizeDelta<storage::replication::CreateDatabaseRpc>(
check_response, main_uuid, std::string(epoch.id()), txn.last_committed_system_timestamp(), txn.timestamp(),
config_);
check_response, epoch.id(), txn.last_committed_system_timestamp(), txn.timestamp(), config_);
}
void PostReplication(replication::RoleMainData &mainData) const override {
@@ -420,10 +413,7 @@ void DbmsHandler::UpdateDurability(const storage::Config &config, std::optional<
#endif
void DbmsHandler::RecoverStorageReplication(DatabaseAccess db_acc, replication::RoleMainData &role_main_data) {
using enum memgraph::flags::Experiments;
auto const is_enterprise = license::global_license_checker.IsEnterpriseValidFast();
auto experimental_system_replication = flags::AreExperimentsEnabled(SYSTEM_REPLICATION);
if ((is_enterprise && experimental_system_replication) || db_acc->name() == dbms::kDefaultDB) {
if (allow_mt_repl || db_acc->name() == dbms::kDefaultDB) {
// Handle global replication state
spdlog::info("Replication configuration will be stored and will be automatically restored in case of a crash.");
// RECOVER REPLICA CONNECTIONS

View File

@@ -29,7 +29,6 @@
#include "kvstore/kvstore.hpp"
#include "license/license.hpp"
#include "replication/replication_client.hpp"
#include "replication_coordination_glue/handler.hpp"
#include "storage/v2/config.hpp"
#include "storage/v2/transaction.hpp"
#include "system/system.hpp"
@@ -107,7 +106,8 @@ class DbmsHandler {
* @param auth pointer to the global authenticator
* @param recovery_on_startup restore databases (and its content) and authentication data
*/
DbmsHandler(storage::Config config, replication::ReplicationState &repl_state, auth::SynchedAuth &auth,
DbmsHandler(storage::Config config, memgraph::system::System &system, replication::ReplicationState &repl_state,
auth::SynchedAuth &auth,
bool recovery_on_startup); // TODO If more arguments are added use a config struct
#else
/**
@@ -115,8 +115,9 @@ class DbmsHandler {
*
* @param configs storage configuration
*/
DbmsHandler(storage::Config config, replication::ReplicationState &repl_state)
DbmsHandler(storage::Config config, memgraph::system::System &system, replication::ReplicationState &repl_state)
: repl_state_{repl_state},
system_{&system},
db_gatekeeper_{[&] {
config.salient.name = kDefaultDB;
return std::move(config);
@@ -260,30 +261,6 @@ class DbmsHandler {
#endif
}
replication::ReplicationState &ReplicationState() { return repl_state_; }
replication::ReplicationState const &ReplicationState() const { return repl_state_; }
bool IsMain() const { return repl_state_.IsMain(); }
bool IsReplica() const { return repl_state_.IsReplica(); }
#ifdef MG_ENTERPRISE
// coordination::CoordinatorState &CoordinatorState() { return coordinator_state_; }
#endif
/**
* @brief Return all active databases.
*
* @return std::vector<std::string>
*/
auto Count() const -> std::size_t {
#ifdef MG_ENTERPRISE
std::shared_lock<LockT> rd(lock_);
return db_handler_.size();
#else
return 1;
#endif
}
/**
* @brief Return the statistics all databases.
*
@@ -599,6 +576,9 @@ class DbmsHandler {
// current replication role. TODO: make Database Access explicit about the role and remove this from
// dbms stuff
replication::ReplicationState &repl_state_; //!< Ref to global replication state
public:
// TODO fix to be non public/remove from dbms....maybe
system::System *system_;
#ifndef MG_ENTERPRISE
mutable utils::Gatekeeper<Database> db_gatekeeper_; //!< Single databases gatekeeper

View File

@@ -144,8 +144,6 @@ class Handler {
auto cbegin() const { return items_.cbegin(); }
auto cend() const { return items_.cend(); }
auto size() const { return items_.size(); }
struct string_hash {
using is_transparent = void;
[[nodiscard]] size_t operator()(const char *s) const { return std::hash<std::string_view>{}(s); }

View File

@@ -19,7 +19,6 @@
#include "storage/v2/durability/durability.hpp"
#include "storage/v2/durability/snapshot.hpp"
#include "storage/v2/durability/version.hpp"
#include "storage/v2/fmt.hpp"
#include "storage/v2/indices/label_index_stats.hpp"
#include "storage/v2/inmemory/storage.hpp"
#include "storage/v2/inmemory/unique_constraints.hpp"
@@ -77,110 +76,58 @@ std::optional<DatabaseAccess> GetDatabaseAccessor(dbms::DbmsHandler *dbms_handle
return std::nullopt;
}
}
void LogWrongMain(const std::optional<utils::UUID> &current_main_uuid, const utils::UUID &main_req_id,
std::string_view rpc_req) {
spdlog::error("Received {} with main_id: {} != current_main_uuid: {}", rpc_req, std::string(main_req_id),
current_main_uuid.has_value() ? std::string(current_main_uuid.value()) : "");
}
} // namespace
void InMemoryReplicationHandlers::Register(dbms::DbmsHandler *dbms_handler, replication::RoleReplicaData &data) {
auto &server = *data.server;
server.rpc_server_.Register<storage::replication::HeartbeatRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received HeartbeatRpc");
InMemoryReplicationHandlers::HeartbeatHandler(dbms_handler, data.uuid_, req_reader, res_builder);
});
void InMemoryReplicationHandlers::Register(dbms::DbmsHandler *dbms_handler, replication::ReplicationServer &server) {
server.rpc_server_.Register<storage::replication::HeartbeatRpc>([dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received HeartbeatRpc");
InMemoryReplicationHandlers::HeartbeatHandler(dbms_handler, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::AppendDeltasRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
[dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received AppendDeltasRpc");
InMemoryReplicationHandlers::AppendDeltasHandler(dbms_handler, data.uuid_, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::SnapshotRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received SnapshotRpc");
InMemoryReplicationHandlers::SnapshotHandler(dbms_handler, data.uuid_, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::WalFilesRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received WalFilesRpc");
InMemoryReplicationHandlers::WalFilesHandler(dbms_handler, data.uuid_, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::CurrentWalRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received CurrentWalRpc");
InMemoryReplicationHandlers::CurrentWalHandler(dbms_handler, data.uuid_, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::TimestampRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received TimestampRpc");
InMemoryReplicationHandlers::TimestampHandler(dbms_handler, data.uuid_, req_reader, res_builder);
});
server.rpc_server_.Register<replication_coordination_glue::SwapMainUUIDRpc>(
[&data, dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received SwapMainUUIDHandler");
InMemoryReplicationHandlers::SwapMainUUIDHandler(dbms_handler, data, req_reader, res_builder);
InMemoryReplicationHandlers::AppendDeltasHandler(dbms_handler, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::SnapshotRpc>([dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received SnapshotRpc");
InMemoryReplicationHandlers::SnapshotHandler(dbms_handler, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::WalFilesRpc>([dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received WalFilesRpc");
InMemoryReplicationHandlers::WalFilesHandler(dbms_handler, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::CurrentWalRpc>([dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received CurrentWalRpc");
InMemoryReplicationHandlers::CurrentWalHandler(dbms_handler, req_reader, res_builder);
});
server.rpc_server_.Register<storage::replication::TimestampRpc>([dbms_handler](auto *req_reader, auto *res_builder) {
spdlog::debug("Received TimestampRpc");
InMemoryReplicationHandlers::TimestampHandler(dbms_handler, req_reader, res_builder);
});
}
void InMemoryReplicationHandlers::SwapMainUUIDHandler(dbms::DbmsHandler *dbms_handler,
replication::RoleReplicaData &role_replica_data,
slk::Reader *req_reader, slk::Builder *res_builder) {
if (!dbms_handler->IsReplica()) {
spdlog::error("Setting main uuid must be performed on replica.");
slk::Save(replication_coordination_glue::SwapMainUUIDRes{false}, res_builder);
return;
}
replication_coordination_glue::SwapMainUUIDReq req;
slk::Load(&req, req_reader);
spdlog::info(fmt::format("Set replica data UUID to main uuid {}", std::string(req.uuid)));
dbms_handler->ReplicationState().TryPersistRoleReplica(role_replica_data.config, req.uuid);
role_replica_data.uuid_ = req.uuid;
slk::Save(replication_coordination_glue::SwapMainUUIDRes{true}, res_builder);
}
void InMemoryReplicationHandlers::HeartbeatHandler(dbms::DbmsHandler *dbms_handler,
const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder) {
void InMemoryReplicationHandlers::HeartbeatHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader,
slk::Builder *res_builder) {
storage::replication::HeartbeatReq req;
slk::Load(&req, req_reader);
auto const db_acc = GetDatabaseAccessor(dbms_handler, req.uuid);
if (!db_acc) {
storage::replication::HeartbeatRes res{false, 0, ""};
slk::Save(res, res_builder);
return;
}
if (!current_main_uuid.has_value() || req.main_uuid != *current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, storage::replication::HeartbeatReq::kType.name);
storage::replication::HeartbeatRes res{false, 0, ""};
slk::Save(res, res_builder);
return;
}
// TODO: this handler is agnostic of InMemory, move to be reused by on-disk
if (!db_acc.has_value()) {
spdlog::warn("No database accessor");
storage::replication::HeartbeatRes res{false, 0, ""};
slk::Save(res, res_builder);
return;
}
auto const *storage = db_acc->get()->storage();
storage::replication::HeartbeatRes res{true, storage->repl_storage_state_.last_commit_timestamp_.load(),
std::string{storage->repl_storage_state_.epoch_.id()}};
slk::Save(res, res_builder);
}
void InMemoryReplicationHandlers::AppendDeltasHandler(dbms::DbmsHandler *dbms_handler,
const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder) {
void InMemoryReplicationHandlers::AppendDeltasHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader,
slk::Builder *res_builder) {
storage::replication::AppendDeltasReq req;
slk::Load(&req, req_reader);
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, storage::replication::AppendDeltasReq::kType.name);
storage::replication::AppendDeltasRes res{false, 0};
slk::Save(res, res_builder);
return;
}
auto db_acc = GetDatabaseAccessor(dbms_handler, req.uuid);
if (!db_acc) {
storage::replication::AppendDeltasRes res{false, 0};
@@ -240,9 +187,8 @@ void InMemoryReplicationHandlers::AppendDeltasHandler(dbms::DbmsHandler *dbms_ha
spdlog::debug("Replication recovery from append deltas finished, replica is now up to date!");
}
void InMemoryReplicationHandlers::SnapshotHandler(dbms::DbmsHandler *dbms_handler,
const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder) {
void InMemoryReplicationHandlers::SnapshotHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader,
slk::Builder *res_builder) {
storage::replication::SnapshotReq req;
slk::Load(&req, req_reader);
auto db_acc = GetDatabaseAccessor(dbms_handler, req.uuid);
@@ -251,12 +197,6 @@ void InMemoryReplicationHandlers::SnapshotHandler(dbms::DbmsHandler *dbms_handle
slk::Save(res, res_builder);
return;
}
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, storage::replication::SnapshotReq::kType.name);
storage::replication::SnapshotRes res{false, 0};
slk::Save(res, res_builder);
return;
}
storage::replication::Decoder decoder(req_reader);
@@ -330,9 +270,8 @@ void InMemoryReplicationHandlers::SnapshotHandler(dbms::DbmsHandler *dbms_handle
spdlog::debug("Replication recovery from snapshot finished!");
}
void InMemoryReplicationHandlers::WalFilesHandler(dbms::DbmsHandler *dbms_handler,
const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder) {
void InMemoryReplicationHandlers::WalFilesHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader,
slk::Builder *res_builder) {
storage::replication::WalFilesReq req;
slk::Load(&req, req_reader);
auto db_acc = GetDatabaseAccessor(dbms_handler, req.uuid);
@@ -341,12 +280,6 @@ void InMemoryReplicationHandlers::WalFilesHandler(dbms::DbmsHandler *dbms_handle
slk::Save(res, res_builder);
return;
}
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, storage::replication::WalFilesReq::kType.name);
storage::replication::WalFilesRes res{false, 0};
slk::Save(res, res_builder);
return;
}
const auto wal_file_number = req.file_number;
spdlog::debug("Received WAL files: {}", wal_file_number);
@@ -365,9 +298,8 @@ void InMemoryReplicationHandlers::WalFilesHandler(dbms::DbmsHandler *dbms_handle
spdlog::debug("Replication recovery from WAL files ended successfully, replica is now up to date!");
}
void InMemoryReplicationHandlers::CurrentWalHandler(dbms::DbmsHandler *dbms_handler,
const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder) {
void InMemoryReplicationHandlers::CurrentWalHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader,
slk::Builder *res_builder) {
storage::replication::CurrentWalReq req;
slk::Load(&req, req_reader);
auto db_acc = GetDatabaseAccessor(dbms_handler, req.uuid);
@@ -377,13 +309,6 @@ void InMemoryReplicationHandlers::CurrentWalHandler(dbms::DbmsHandler *dbms_hand
return;
}
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, storage::replication::CurrentWalReq::kType.name);
storage::replication::CurrentWalRes res{false, 0};
slk::Save(res, res_builder);
return;
}
storage::replication::Decoder decoder(req_reader);
auto *storage = static_cast<storage::InMemoryStorage *>(db_acc->get()->storage());
@@ -445,9 +370,8 @@ void InMemoryReplicationHandlers::LoadWal(storage::InMemoryStorage *storage, sto
}
}
void InMemoryReplicationHandlers::TimestampHandler(dbms::DbmsHandler *dbms_handler,
const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder) {
void InMemoryReplicationHandlers::TimestampHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader,
slk::Builder *res_builder) {
storage::replication::TimestampReq req;
slk::Load(&req, req_reader);
auto const db_acc = GetDatabaseAccessor(dbms_handler, req.uuid);
@@ -457,13 +381,6 @@ void InMemoryReplicationHandlers::TimestampHandler(dbms::DbmsHandler *dbms_handl
return;
}
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, storage::replication::TimestampReq::kType.name);
storage::replication::CurrentWalRes res{false, 0};
slk::Save(res, res_builder);
return;
}
// TODO: this handler is agnostic of InMemory, move to be reused by on-disk
auto const *storage = db_acc->get()->storage();
storage::replication::TimestampRes res{true, storage->repl_storage_state_.last_commit_timestamp_.load()};

View File

@@ -12,7 +12,6 @@
#pragma once
#include "replication/replication_server.hpp"
#include "replication/state.hpp"
#include "storage/v2/replication/serialization.hpp"
namespace memgraph::storage {
@@ -24,30 +23,21 @@ class DbmsHandler;
class InMemoryReplicationHandlers {
public:
static void Register(dbms::DbmsHandler *dbms_handler, replication::RoleReplicaData &data);
static void Register(dbms::DbmsHandler *dbms_handler, replication::ReplicationServer &server);
private:
// RPC handlers
static void HeartbeatHandler(dbms::DbmsHandler *dbms_handler, const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder);
static void HeartbeatHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
static void AppendDeltasHandler(dbms::DbmsHandler *dbms_handler, const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder);
static void AppendDeltasHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
static void SnapshotHandler(dbms::DbmsHandler *dbms_handler, const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder);
static void SnapshotHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
static void WalFilesHandler(dbms::DbmsHandler *dbms_handler, const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder);
static void WalFilesHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
static void CurrentWalHandler(dbms::DbmsHandler *dbms_handler, const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder);
static void CurrentWalHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
static void TimestampHandler(dbms::DbmsHandler *dbms_handler, const std::optional<utils::UUID> &current_main_uuid,
slk::Reader *req_reader, slk::Builder *res_builder);
static void SwapMainUUIDHandler(dbms::DbmsHandler *dbms_handler, replication::RoleReplicaData &role_replica_data,
slk::Reader *req_reader, slk::Builder *res_builder);
static void TimestampHandler(dbms::DbmsHandler *dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
static void LoadWal(storage::InMemoryStorage *storage, storage::replication::Decoder *decoder);

View File

@@ -21,8 +21,7 @@ namespace memgraph::dbms {
#ifdef MG_ENTERPRISE
void CreateDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, DbmsHandler &dbms_handler,
slk::Reader *req_reader, slk::Builder *res_builder) {
DbmsHandler &dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder) {
using memgraph::storage::replication::CreateDatabaseRes;
CreateDatabaseRes res(CreateDatabaseRes::Result::FAILURE);
@@ -36,12 +35,6 @@ void CreateDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system
memgraph::storage::replication::CreateDatabaseReq req;
memgraph::slk::Load(&req, req_reader);
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, memgraph::storage::replication::CreateDatabaseReq::kType.name);
memgraph::slk::Save(res, res_builder);
return;
}
// Note: No need to check epoch, recovery mechanism is done by a full uptodate snapshot
// of the set of databases. Hence no history exists to maintain regarding epoch change.
// If MAIN has changed we need to check this new group_timestamp is consistent with
@@ -70,8 +63,7 @@ void CreateDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system
memgraph::slk::Save(res, res_builder);
}
void DropDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, DbmsHandler &dbms_handler,
void DropDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access, DbmsHandler &dbms_handler,
slk::Reader *req_reader, slk::Builder *res_builder) {
using memgraph::storage::replication::DropDatabaseRes;
DropDatabaseRes res(DropDatabaseRes::Result::FAILURE);
@@ -86,12 +78,6 @@ void DropDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_s
memgraph::storage::replication::DropDatabaseReq req;
memgraph::slk::Load(&req, req_reader);
if (!current_main_uuid.has_value() || req.main_uuid != current_main_uuid) [[unlikely]] {
LogWrongMain(current_main_uuid, req.main_uuid, memgraph::storage::replication::DropDatabaseReq::kType.name);
memgraph::slk::Save(res, res_builder);
return;
}
// Note: No need to check epoch, recovery mechanism is done by a full uptodate snapshot
// of the set of databases. Hence no history exists to maintain regarding epoch change.
// If MAIN has changed we need to check this new group_timestamp is consistent with
@@ -191,14 +177,14 @@ void Register(replication::RoleReplicaData const &data, system::ReplicaHandlerAc
dbms::DbmsHandler &dbms_handler) {
// NOTE: Register even without license as the user could add a license at run-time
data.server->rpc_server_.Register<storage::replication::CreateDatabaseRpc>(
[&data, system_state_access, &dbms_handler](auto *req_reader, auto *res_builder) mutable {
[system_state_access, &dbms_handler](auto *req_reader, auto *res_builder) mutable {
spdlog::debug("Received CreateDatabaseRpc");
CreateDatabaseHandler(system_state_access, data.uuid_, dbms_handler, req_reader, res_builder);
CreateDatabaseHandler(system_state_access, dbms_handler, req_reader, res_builder);
});
data.server->rpc_server_.Register<storage::replication::DropDatabaseRpc>(
[&data, system_state_access, &dbms_handler](auto *req_reader, auto *res_builder) mutable {
[system_state_access, &dbms_handler](auto *req_reader, auto *res_builder) mutable {
spdlog::debug("Received DropDatabaseRpc");
DropDatabaseHandler(system_state_access, data.uuid_, dbms_handler, req_reader, res_builder);
DropDatabaseHandler(system_state_access, dbms_handler, req_reader, res_builder);
});
}
#endif

View File

@@ -17,21 +17,11 @@
#include "system/state.hpp"
namespace memgraph::dbms {
#ifdef MG_ENTERPRISE
inline void LogWrongMain(const std::optional<utils::UUID> &current_main_uuid, const utils::UUID &main_req_id,
std::string_view rpc_req) {
spdlog::error("Received {} with main_id: {} != current_main_uuid: {}", rpc_req, std::string(main_req_id),
current_main_uuid.has_value() ? std::string(current_main_uuid.value()) : "");
}
// RPC handlers
void CreateDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, DbmsHandler &dbms_handler,
slk::Reader *req_reader, slk::Builder *res_builder);
void DropDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access,
const std::optional<utils::UUID> &current_main_uuid, DbmsHandler &dbms_handler,
DbmsHandler &dbms_handler, slk::Reader *req_reader, slk::Builder *res_builder);
void DropDatabaseHandler(memgraph::system::ReplicaHandlerAccessToState &system_state_access, DbmsHandler &dbms_handler,
slk::Reader *req_reader, slk::Builder *res_builder);
bool SystemRecoveryHandler(DbmsHandler &dbms_handler, const std::vector<storage::SalientConfig> &database_configs);

View File

@@ -57,7 +57,6 @@ namespace slk {
// Serialize code for CreateDatabaseReq
void Save(const memgraph::storage::replication::CreateDatabaseReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.main_uuid, builder);
memgraph::slk::Save(self.epoch_id, builder);
memgraph::slk::Save(self.expected_group_timestamp, builder);
memgraph::slk::Save(self.new_group_timestamp, builder);
@@ -65,7 +64,6 @@ void Save(const memgraph::storage::replication::CreateDatabaseReq &self, memgrap
}
void Load(memgraph::storage::replication::CreateDatabaseReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->main_uuid, reader);
memgraph::slk::Load(&self->epoch_id, reader);
memgraph::slk::Load(&self->expected_group_timestamp, reader);
memgraph::slk::Load(&self->new_group_timestamp, reader);
@@ -89,7 +87,6 @@ void Load(memgraph::storage::replication::CreateDatabaseRes *self, memgraph::slk
// Serialize code for DropDatabaseReq
void Save(const memgraph::storage::replication::DropDatabaseReq &self, memgraph::slk::Builder *builder) {
memgraph::slk::Save(self.main_uuid, builder);
memgraph::slk::Save(self.epoch_id, builder);
memgraph::slk::Save(self.expected_group_timestamp, builder);
memgraph::slk::Save(self.new_group_timestamp, builder);
@@ -97,7 +94,6 @@ void Save(const memgraph::storage::replication::DropDatabaseReq &self, memgraph:
}
void Load(memgraph::storage::replication::DropDatabaseReq *self, memgraph::slk::Reader *reader) {
memgraph::slk::Load(&self->main_uuid, reader);
memgraph::slk::Load(&self->epoch_id, reader);
memgraph::slk::Load(&self->expected_group_timestamp, reader);
memgraph::slk::Load(&self->new_group_timestamp, reader);

View File

@@ -29,15 +29,13 @@ struct CreateDatabaseReq {
static void Load(CreateDatabaseReq *self, memgraph::slk::Reader *reader);
static void Save(const CreateDatabaseReq &self, memgraph::slk::Builder *builder);
CreateDatabaseReq() = default;
CreateDatabaseReq(const utils::UUID &main_uuid, std::string epoch_id, uint64_t expected_group_timestamp,
uint64_t new_group_timestamp, storage::SalientConfig config)
: main_uuid(main_uuid),
epoch_id(std::move(epoch_id)),
CreateDatabaseReq(std::string_view epoch_id, uint64_t expected_group_timestamp, uint64_t new_group_timestamp,
storage::SalientConfig config)
: epoch_id(std::string(epoch_id)),
expected_group_timestamp{expected_group_timestamp},
new_group_timestamp(new_group_timestamp),
config(std::move(config)) {}
utils::UUID main_uuid;
std::string epoch_id;
uint64_t expected_group_timestamp;
uint64_t new_group_timestamp;
@@ -67,15 +65,13 @@ struct DropDatabaseReq {
static void Load(DropDatabaseReq *self, memgraph::slk::Reader *reader);
static void Save(const DropDatabaseReq &self, memgraph::slk::Builder *builder);
DropDatabaseReq() = default;
DropDatabaseReq(const utils::UUID &main_uuid, std::string epoch_id, uint64_t expected_group_timestamp,
uint64_t new_group_timestamp, const utils::UUID &uuid)
: main_uuid(main_uuid),
epoch_id(std::move(epoch_id)),
DropDatabaseReq(std::string_view epoch_id, uint64_t expected_group_timestamp, uint64_t new_group_timestamp,
const utils::UUID &uuid)
: epoch_id(std::string(epoch_id)),
expected_group_timestamp{expected_group_timestamp},
new_group_timestamp(new_group_timestamp),
uuid(uuid) {}
utils::UUID main_uuid;
std::string epoch_id;
uint64_t expected_group_timestamp;
uint64_t new_group_timestamp;

View File

@@ -10,7 +10,6 @@
// licenses/APL.txt.
#pragma once
#include <algorithm>
#include <atomic>
#include <compare>
#include <cstdint>

View File

@@ -1,17 +1,12 @@
add_library(mg-flags STATIC
audit.cpp
bolt.cpp
general.cpp
isolation_level.cpp
log_level.cpp
memory_limit.cpp
run_time_configurable.cpp
storage_mode.cpp
query.cpp
replication.cpp
experimental.cpp
experimental.hpp)
target_include_directories(mg-flags PUBLIC include)
target_link_libraries(mg-flags
PUBLIC spdlog::spdlog mg-settings mg-utils
PRIVATE lib::rangev3)
add_library(mg-flags STATIC audit.cpp
bolt.cpp
general.cpp
isolation_level.cpp
log_level.cpp
memory_limit.cpp
run_time_configurable.cpp
storage_mode.cpp
query.cpp
replication.cpp)
target_include_directories(mg-flags PUBLIC ${CMAKE_SOURCE_DIR}/include)
target_link_libraries(mg-flags PUBLIC spdlog::spdlog mg-settings mg-utils)

View File

@@ -12,7 +12,6 @@
#include "flags/audit.hpp"
#include "flags/bolt.hpp"
#include "flags/experimental.hpp"
#include "flags/general.hpp"
#include "flags/isolation_level.hpp"
#include "flags/log_level.hpp"

View File

@@ -1,68 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#include "flags/experimental.hpp"
#include "range/v3/all.hpp"
#include "utils/string.hpp"
#include <map>
#include <string_view>
// Bolt server flags.
// NOLINTNEXTLINE (cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_string(experimental_enabled, "",
"Experimental features to be used, comma seperated. Options [system-replication, high-availability]");
using namespace std::string_view_literals;
namespace memgraph::flags {
auto const mapping = std::map{std::pair{"system-replication"sv, Experiments::SYSTEM_REPLICATION},
std::pair{"high-availability"sv, Experiments::HIGH_AVAILABILITY}};
auto ExperimentsInstance() -> Experiments & {
static auto instance = Experiments{};
return instance;
}
bool AreExperimentsEnabled(Experiments experiments) {
using t = std::underlying_type_t<Experiments>;
auto actual = static_cast<t>(ExperimentsInstance());
auto check = static_cast<t>(experiments);
return (actual & check) == check;
}
void InitializeExperimental() {
namespace rv = ranges::views;
auto const connonicalize_string = [](auto &&rng) {
auto const is_space = [](auto c) { return c == ' '; };
auto const to_lower = [](unsigned char c) { return std::tolower(c); };
return rng | rv::drop_while(is_space) | rv::take_while(std::not_fn(is_space)) | rv::transform(to_lower) |
ranges::to<std::string>;
};
auto const mapping_end = mapping.cend();
using underlying_type = std::underlying_type_t<Experiments>;
auto to_set = underlying_type{};
for (auto &&experiment : FLAGS_experimental_enabled | rv::split(',') | rv::transform(connonicalize_string)) {
if (auto it = mapping.find(experiment); it != mapping_end) {
to_set |= static_cast<underlying_type>(it->second);
}
}
ExperimentsInstance() = static_cast<Experiments>(to_set);
}
} // namespace memgraph::flags

View File

@@ -1,33 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#include "gflags/gflags.h"
// Short help flag.
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_string(experimental_enabled);
namespace memgraph::flags {
// Each bit is an enabled experiment
// old experiments can be reused once code cleanup has happened
enum class Experiments : uint8_t {
SYSTEM_REPLICATION = 1 << 0,
HIGH_AVAILABILITY = 1 << 1,
};
bool AreExperimentsEnabled(Experiments experiments);
void InitializeExperimental();
} // namespace memgraph::flags

View File

@@ -13,17 +13,9 @@
#ifdef MG_ENTERPRISE
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_bool(coordinator, false, "Controls whether the instance is a replication coordinator.");
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_uint32(coordinator_server_port, 0, "Port on which coordinator servers will be started.");
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_uint32(raft_server_port, 0, "Port on which raft servers will be started.");
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_uint32(raft_server_id, 0, "Unique ID of the raft server.");
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_uint32(instance_down_timeout_sec, 5, "Time duration after which an instance is considered down.");
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_uint32(instance_health_check_frequency_sec, 1, "The time duration between two health checks/pings.");
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DEFINE_uint32(instance_get_uuid_frequency_sec, 10, "The time duration between two instance uuid checks.");
#endif
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)

View File

@@ -15,17 +15,9 @@
#ifdef MG_ENTERPRISE
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_bool(coordinator);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_uint32(coordinator_server_port);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_uint32(raft_server_port);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_uint32(raft_server_id);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_uint32(instance_down_timeout_sec);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_uint32(instance_health_check_frequency_sec);
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
DECLARE_uint32(instance_get_uuid_frequency_sec);
#endif
// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -22,7 +22,6 @@
#include "integrations/constants.hpp"
#include "integrations/kafka/exceptions.hpp"
#include "integrations/kafka/fmt.hpp"
#include "utils/exceptions.hpp"
#include "utils/logging.hpp"
#include "utils/on_scope_exit.hpp"

View File

@@ -1,25 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#if FMT_VERSION > 90000
#include <fmt/ostream.h>
#include <librdkafka/rdkafkacpp.h>
inline std::ostream &operator<<(std::ostream &os, const RdKafka::ErrorCode &code) {
os << RdKafka::err2str(code);
return os;
}
template <>
class fmt::formatter<RdKafka::ErrorCode> : public fmt::ostream_formatter {};
#endif

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -15,12 +15,12 @@
#include <chrono>
#include <thread>
#include <fmt/format.h>
#include <pulsar/Client.h>
#include <pulsar/InitialPosition.h>
#include "integrations/constants.hpp"
#include "integrations/pulsar/exceptions.hpp"
#include "integrations/pulsar/fmt.hpp"
#include "utils/concepts.hpp"
#include "utils/logging.hpp"
#include "utils/on_scope_exit.hpp"

View File

@@ -1,21 +0,0 @@
// Copyright 2024 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
// License, and you may not use this file except in compliance with the Business Source License.
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0, included in the file
// licenses/APL.txt.
#pragma once
#if FMT_VERSION > 90000
#include <fmt/ostream.h>
#include "integrations/pulsar/consumer.hpp"
template <>
class fmt::formatter<memgraph::integrations::pulsar::pulsar_client::Result> : public fmt::ostream_formatter {};
#endif

View File

@@ -39,7 +39,7 @@ Endpoint::IpFamily Endpoint::GetIpFamily(const std::string &address) {
}
std::optional<std::pair<std::string, uint16_t>> Endpoint::ParseSocketOrIpAddress(
const std::string &address, const std::optional<uint16_t> default_port) {
const std::string &address, const std::optional<uint16_t> default_port = {}) {
/// expected address format:
/// - "ip_address:port_number"
/// - "ip_address"

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -61,8 +61,8 @@ struct Endpoint {
* it into an ip address and a port number; even if a default port is given,
* it won't be used, as we expect that it is given in the address string.
*/
static std::optional<std::pair<std::string, uint16_t>> ParseSocketOrIpAddress(
const std::string &address, std::optional<uint16_t> default_port = {});
static std::optional<std::pair<std::string, uint16_t>> ParseSocketOrIpAddress(const std::string &address,
std::optional<uint16_t> default_port);
/**
* Tries to parse given string as either socket address or hostname.

View File

@@ -1,4 +1,4 @@
// Copyright 2023 Memgraph Ltd.
// Copyright 2022 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -11,7 +11,6 @@
#pragma once
#include <cstddef>
#include <cstdint>
namespace memgraph::io::network {

View File

@@ -160,14 +160,13 @@ class KVStore final {
* and behaves as if all of those pairs are stored in a single iterable
* collection of std::pair<std::string, std::string>.
*/
class iterator final {
class iterator final : public std::iterator<std::input_iterator_tag, // iterator_category
std::pair<std::string, std::string>, // value_type
long, // difference_type
const std::pair<std::string, std::string> *, // pointer
const std::pair<std::string, std::string> & // reference
> {
public:
using iterator_concept [[maybe_unused]] = std::input_iterator_tag;
using value_type = std::pair<std::string, std::string>;
using difference_type = long;
using pointer = const std::pair<std::string, std::string> *;
using reference = const std::pair<std::string, std::string> &;
explicit iterator(const KVStore *kvstore, const std::string &prefix = "", bool at_end = false);
iterator(const iterator &other) = delete;

View File

@@ -134,7 +134,6 @@ int main(int argc, char **argv) {
}
memgraph::flags::InitializeLogger();
memgraph::flags::InitializeExperimental();
// Unhandled exception handler init.
std::set_terminate(&memgraph::utils::TerminateHandler);
@@ -356,11 +355,6 @@ int main(int argc, char **argv) {
memgraph::query::InterpreterConfig interp_config{
.query = {.allow_load_csv = FLAGS_allow_load_csv},
.replication_replica_check_frequency = std::chrono::seconds(FLAGS_replication_replica_check_frequency_sec),
#ifdef MG_ENTERPRISE
.instance_down_timeout_sec = std::chrono::seconds(FLAGS_instance_down_timeout_sec),
.instance_health_check_frequency_sec = std::chrono::seconds(FLAGS_instance_health_check_frequency_sec),
.instance_get_uuid_frequency_sec = std::chrono::seconds(FLAGS_instance_get_uuid_frequency_sec),
#endif
.default_kafka_bootstrap_servers = FLAGS_kafka_bootstrap_servers,
.default_pulsar_service_url = FLAGS_pulsar_service_url,
.stream_transaction_conflict_retries = FLAGS_stream_transaction_conflict_retries,
@@ -402,7 +396,7 @@ int main(int argc, char **argv) {
memgraph::coordination::CoordinatorState coordinator_state;
#endif
memgraph::dbms::DbmsHandler dbms_handler(db_config, repl_state
memgraph::dbms::DbmsHandler dbms_handler(db_config, system, repl_state
#ifdef MG_ENTERPRISE
,
auth_, FLAGS_data_recovery_on_startup
@@ -415,7 +409,7 @@ int main(int argc, char **argv) {
auto replication_handler = memgraph::replication::ReplicationHandler{repl_state, dbms_handler
#ifdef MG_ENTERPRISE
,
system, auth_
&system, auth_
#endif
};

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -60,13 +60,10 @@ void *my_alloc(extent_hooks_t *extent_hooks, void *new_addr, size_t size, size_t
unsigned arena_ind) {
// This needs to be before, to throw exception in case of too big alloc
if (*commit) [[likely]] {
memgraph::utils::total_memory_tracker.Alloc(static_cast<int64_t>(size));
if (GetQueriesMemoryControl().IsThreadTracked()) [[unlikely]] {
bool ok = GetQueriesMemoryControl().TrackAllocOnCurrentThread(size);
if (!ok) return nullptr;
GetQueriesMemoryControl().TrackAllocOnCurrentThread(size);
}
// This needs to be here so it doesn't get incremented in case the first TrackAlloc throws an exception
bool ok = memgraph::utils::total_memory_tracker.Alloc(static_cast<int64_t>(size));
if (!ok) return nullptr;
}
auto *ptr = old_hooks->alloc(extent_hooks, new_addr, size, alignment, zero, commit, arena_ind);
@@ -120,15 +117,11 @@ static bool my_commit(extent_hooks_t *extent_hooks, void *addr, size_t size, siz
return err;
}
[[maybe_unused]] auto blocker = memgraph::utils::MemoryTracker::OutOfMemoryExceptionBlocker{};
memgraph::utils::total_memory_tracker.Alloc(static_cast<int64_t>(length));
if (GetQueriesMemoryControl().IsThreadTracked()) [[unlikely]] {
bool ok = GetQueriesMemoryControl().TrackAllocOnCurrentThread(length);
DMG_ASSERT(ok);
GetQueriesMemoryControl().TrackAllocOnCurrentThread(size);
}
auto ok = memgraph::utils::total_memory_tracker.Alloc(static_cast<int64_t>(length));
DMG_ASSERT(ok);
return false;
}

View File

@@ -1,4 +1,4 @@
// Copyright 2024 Memgraph Ltd.
// Copyright 2023 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -28,12 +28,6 @@ void *newImpl(const std::size_t size) {
return ptr;
}
[[maybe_unused]] auto blocker = memgraph::utils::MemoryTracker::OutOfMemoryExceptionBlocker{};
auto maybe_msg = memgraph::utils::MemoryErrorStatus().msg();
if (maybe_msg) {
throw memgraph::utils::OutOfMemoryException{std::move(*maybe_msg)};
}
throw std::bad_alloc{};
}
@@ -43,21 +37,11 @@ void *newImpl(const std::size_t size, const std::align_val_t align) {
return ptr;
}
[[maybe_unused]] auto blocker = memgraph::utils::MemoryTracker::OutOfMemoryExceptionBlocker{};
auto maybe_msg = memgraph::utils::MemoryErrorStatus().msg();
if (maybe_msg) {
throw memgraph::utils::OutOfMemoryException{std::move(*maybe_msg)};
}
throw std::bad_alloc{};
}
void *newNoExcept(const std::size_t size) noexcept {
[[maybe_unused]] auto blocker = memgraph::utils::MemoryTracker::OutOfMemoryExceptionBlocker{};
return malloc(size);
}
void *newNoExcept(const std::size_t size) noexcept { return malloc(size); }
void *newNoExcept(const std::size_t size, const std::align_val_t align) noexcept {
[[maybe_unused]] auto blocker = memgraph::utils::MemoryTracker::OutOfMemoryExceptionBlocker{};
return aligned_alloc(size, static_cast<std::size_t>(align));
}

View File

@@ -54,14 +54,14 @@ void QueriesMemoryControl::EraseThreadToTransactionId(const std::thread::id &thr
}
}
bool QueriesMemoryControl::TrackAllocOnCurrentThread(size_t size) {
void QueriesMemoryControl::TrackAllocOnCurrentThread(size_t size) {
auto thread_id_to_transaction_id_accessor = thread_id_to_transaction_id.access();
// we might be just constructing mapping between thread id and transaction id
// so we miss this allocation
auto thread_id_to_transaction_id_elem = thread_id_to_transaction_id_accessor.find(std::this_thread::get_id());
if (thread_id_to_transaction_id_elem == thread_id_to_transaction_id_accessor.end()) {
return true;
return;
}
auto transaction_id_to_tracker_accessor = transaction_id_to_tracker.access();
@@ -71,10 +71,10 @@ bool QueriesMemoryControl::TrackAllocOnCurrentThread(size_t size) {
// It can happen that some allocation happens between mapping thread to
// transaction id, so we miss this allocation
if (transaction_id_to_tracker == transaction_id_to_tracker_accessor.end()) [[unlikely]] {
return true;
return;
}
auto &query_tracker = transaction_id_to_tracker->tracker;
return query_tracker.TrackAlloc(size);
query_tracker.TrackAlloc(size);
}
void QueriesMemoryControl::TrackFreeOnCurrentThread(size_t size) {
@@ -110,7 +110,7 @@ void QueriesMemoryControl::CreateTransactionIdTracker(uint64_t transaction_id, s
bool QueriesMemoryControl::EraseTransactionIdTracker(uint64_t transaction_id) {
auto transaction_id_to_tracker_accessor = transaction_id_to_tracker.access();
auto removed = transaction_id_to_tracker_accessor.remove(transaction_id);
auto removed = transaction_id_to_tracker.access().remove(transaction_id);
return removed;
}

View File

@@ -62,7 +62,7 @@ class QueriesMemoryControl {
// Find tracker for current thread if exists, track
// query allocation and procedure allocation if
// necessary
bool TrackAllocOnCurrentThread(size_t size);
void TrackAllocOnCurrentThread(size_t size);
// Find tracker for current thread if exists, track
// query allocation and procedure allocation if

View File

@@ -139,11 +139,6 @@ struct NodeId {
std::string id_space;
};
#if FMT_VERSION > 90000
template <>
class fmt::formatter<NodeId> : public fmt::ostream_formatter {};
#endif
bool operator==(const NodeId &a, const NodeId &b) { return a.id == b.id && a.id_space == b.id_space; }
std::ostream &operator<<(std::ostream &stream, const NodeId &node_id) {

View File

@@ -1,4 +1,4 @@
// Copyright 2023 Memgraph Ltd.
// Copyright 2022 Memgraph Ltd.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.txt; by using this file, you agree to be bound by the terms of the Business Source
@@ -274,8 +274,3 @@ inline void RestoreError(ExceptionInfo exc_info) {
}
} // namespace memgraph::py
#if FMT_VERSION > 90000
template <>
class fmt::formatter<memgraph::py::ExceptionInfo> : public fmt::ostream_formatter {};
#endif

View File

@@ -19,7 +19,6 @@
#include "query/db_accessor.hpp"
#include "query/exceptions.hpp"
#include "query/fmt.hpp"
#include "query/frontend/ast/ast.hpp"
#include "query/frontend/semantic/symbol.hpp"
#include "query/typed_value.hpp"

View File

@@ -22,10 +22,6 @@ struct InterpreterConfig {
// The same as \ref memgraph::replication::ReplicationClientConfig
std::chrono::seconds replication_replica_check_frequency{1};
std::chrono::seconds instance_down_timeout_sec{5};
std::chrono::seconds instance_health_check_frequency_sec{1};
std::chrono::seconds instance_get_uuid_frequency_sec{10};
std::string default_kafka_bootstrap_servers;
std::string default_pulsar_service_url;
uint32_t stream_transaction_conflict_retries;

View File

@@ -54,10 +54,6 @@ class EdgeAccessor final {
return impl_.GetProperty(key, view);
}
storage::Result<uint64_t> GetPropertySize(storage::PropertyId key, storage::View view) const {
return impl_.GetPropertySize(key, view);
}
storage::Result<storage::PropertyValue> SetProperty(storage::PropertyId key, const storage::PropertyValue &value) {
return impl_.SetProperty(key, value);
}
@@ -133,10 +129,6 @@ class VertexAccessor final {
return impl_.GetProperty(key, view);
}
storage::Result<uint64_t> GetPropertySize(storage::PropertyId key, storage::View view) const {
return impl_.GetPropertySize(key, view);
}
storage::Result<storage::PropertyValue> SetProperty(storage::PropertyId key, const storage::PropertyValue &value) {
return impl_.SetProperty(key, value);
}
@@ -276,10 +268,6 @@ class SubgraphVertexAccessor final {
return impl_.GetProperty(view, key);
}
storage::Result<uint64_t> GetPropertySize(storage::PropertyId key, storage::View view) const {
return impl_.GetPropertySize(key, view);
}
storage::Gid Gid() const noexcept { return impl_.Gid(); }
storage::Result<size_t> InDegree(storage::View view) const { return impl_.InDegree(view); }
@@ -541,10 +529,6 @@ class DbAccessor final {
storage::PropertyId NameToProperty(const std::string_view name) { return accessor_->NameToProperty(name); }
std::optional<storage::PropertyId> NameToPropertyIfExists(std::string_view name) const {
return accessor_->NameToPropertyIfExists(name);
}
storage::LabelId NameToLabel(const std::string_view name) { return accessor_->NameToLabel(name); }
storage::EdgeTypeId NameToEdgeType(const std::string_view name) { return accessor_->NameToEdgeType(name); }

Some files were not shown because too many files have changed in this diff Show More