Skip to content

ClickHouse Migration Cluster Check on 8101/merge by @andriidudar #932

ClickHouse Migration Cluster Check on 8101/merge by @andriidudar

ClickHouse Migration Cluster Check on 8101/merge by @andriidudar #932

name: "ClickHouse Migration Cluster Check"
run-name: "ClickHouse Migration Cluster Check on ${{ github.ref_name }} by @${{ github.actor }}"
permissions:
contents: read
on:
pull_request:
paths:
- "apps/opik-backend/src/main/resources/liquibase/db-app-analytics/migrations/**/*.sql"
push:
branches:
- "main"
paths:
- "apps/opik-backend/src/main/resources/liquibase/db-app-analytics/migrations/**/*.sql"
workflow_dispatch:
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
cancel-in-progress: ${{ github.event_name == 'pull_request' }}
jobs:
validate-clickhouse-migrations:
runs-on: ubuntu-latest
timeout-minutes: 5
steps:
- name: Checkout
uses: actions/checkout@v7
with:
fetch-depth: 1
- name: Get changed ClickHouse migration files
id: changed-files
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
echo "🔍 Detecting ClickHouse migration files that were added or modified..."
if [ "${{ github.event_name }}" = "pull_request" ]; then
# For PRs, use GitHub's built-in files API
gh api repos/${{ github.repository }}/pulls/${{ github.event.number }}/files \
--jq '.[] | select(.status == "added" or .status == "modified") | select(.filename | test("^apps/opik-backend/src/main/resources/liquibase/db-app-analytics/migrations/.*\\.sql$")) | .filename' > changed_files.txt
CHANGED_FILES=$(cat changed_files.txt || echo "")
rm -f changed_files.txt
else
# For pushes to main, use GitHub API to get commit files
# This works reliably regardless of fetch-depth
gh api repos/${{ github.repository }}/commits/${{ github.sha }} \
--jq '.files[] | select(.status == "added" or .status == "modified") | select(.filename | test("^apps/opik-backend/src/main/resources/liquibase/db-app-analytics/migrations/.*\\.sql$")) | .filename' > changed_files.txt
CHANGED_FILES=$(cat changed_files.txt || echo "")
rm -f changed_files.txt
fi
if [ -n "$CHANGED_FILES" ]; then
echo "📄 Found ClickHouse migration files to validate:"
# shellcheck disable=SC2086 # intentional word-split on newlines in CHANGED_FILES
printf '%s\n' $CHANGED_FILES | while read -r file; do
[ -n "$file" ] && echo " - $file"
done
# Save the files to environment variable for next step
echo "MIGRATION_FILES<<EOF" >> "$GITHUB_ENV"
# shellcheck disable=SC2086 # intentional word-split on newlines in CHANGED_FILES
printf '%s\n' $CHANGED_FILES | while read -r file; do
[ -n "$file" ] && echo "$file" >> "$GITHUB_ENV"
done
echo "EOF" >> "$GITHUB_ENV"
echo "HAS_MIGRATION_FILES=true" >> "$GITHUB_ENV"
else
echo "📋 No ClickHouse migration files were added or modified."
echo "✅ No validation needed."
echo "HAS_MIGRATION_FILES=false" >> "$GITHUB_ENV"
fi
- name: Validate ClickHouse migrations have ON CLUSTER clause
if: env.HAS_MIGRATION_FILES == 'true'
run: |
echo "🔍 Validating ClickHouse migrations for proper ON CLUSTER clause usage..."
echo "📖 Reference: https://clickhouse.com/docs/sql-reference/distributed-ddl"
echo
# Run the validation script with the migration files
# Use xargs with -0 flag and tr to handle newline-separated list safely
# This approach correctly handles filenames with spaces across different platforms
if echo "$MIGRATION_FILES" | tr '\n' '\0' | xargs -0 ./scripts/check_clickhouse_migrations_cluster.sh; then
echo "✅ All modified ClickHouse migrations are properly configured!"
echo "🎉 DDL operations will be executed distributively across the cluster."
else
echo "❌ ClickHouse migration validation failed!"
echo
echo "💡 What this means:"
echo " - DDL operations without ON CLUSTER clause will only execute on a single node"
echo " - This can cause inconsistencies in distributed ClickHouse deployments"
echo " - All CREATE, DROP, ALTER, RENAME, and EXCHANGE statements must include ON CLUSTER '{cluster}'"
echo
echo "🔧 How to fix:"
echo " 1. Add 'ON CLUSTER '\"'\"'{cluster}'\"'\"'' to all DDL statements in your migration files"
echo " 2. Example: CREATE TABLE my_table ON CLUSTER '{cluster}' (...)"
echo " 3. Reference: https://clickhouse.com/docs/sql-reference/distributed-ddl"
echo
exit 1
fi
- name: Summary for no migration files
if: env.HAS_MIGRATION_FILES == 'false'
run: |
echo "=========================================="
echo "✅ No ClickHouse Migration Files Changed"
echo "=========================================="
echo "📋 No ClickHouse migration files were added or modified in this change."
echo "✅ No validation required."
echo
echo "💡 Note: This check only validates NEW or MODIFIED migration files."
echo " Existing migration files are not checked as they cannot be changed."
- name: Summary for validated files
if: env.HAS_MIGRATION_FILES == 'true' && success()
run: |
echo "=========================================="
echo "✅ ClickHouse Migration Validation PASSED"
echo "=========================================="
echo "✅ All modified DDL operations include proper ON CLUSTER clause"
echo "✅ Migrations are ready for distributed ClickHouse deployment"
echo "✅ No cluster inconsistencies will occur"
echo
echo "🎯 This check ensures that:"
echo " • CREATE statements will create tables/indexes on all cluster nodes"
echo " • DROP statements will remove objects from all cluster nodes"
echo " • ALTER statements will modify schema on all cluster nodes"
echo " • RENAME statements will rename objects on all cluster nodes"
echo
echo "📚 For more information about distributed DDL:"
echo " https://clickhouse.com/docs/sql-reference/distributed-ddl"