Aias00 opened a new issue, #6962: URL: https://github.com/apache/rocketmq-dashboard/issues/6962
## Failure Backend Tests for #6954 (documentation-only, base `rocketmq-studio@c34535db0`) failed in `ResourceOwnershipGuardTest.migrationConcurrentStartupIsIdempotentTest`: ``` Index "uk_group_name" already exists CREATE UNIQUE INDEX uk_group_name ON rmq_instance_group (name) [42111-240] ResourceOwnershipSchemaMigration.ensureIndex:132 ``` Run: https://github.com/apache/rocketmq-dashboard/actions/runs/39145822282/job/117729324192 ## Remaining boundary `ensureIndex` handles a racing CREATE by checking the metadata once and accepting only the identical index definition. H2 can expose an incomplete metadata view while the competing DDL is published, without throwing the `ConcurrentModificationException` covered by merged #6736 / #6022. A single nonmatching read rethrows the CREATE error and makes startup/CI intermittent. ## Proposed repair and verification Boundedly recheck the expected index definition after failed CREATE, allowing the competing metadata publication to settle. Accept only the exact expected uniqueness and columns. Preserve the original SQL error if the definition never matches; preserve thread interruption; do not serialize the migrations or weaken the original concurrent-startup test. Add a deterministic JDBC-boundary test for delayed metadata visibility with a real H2-created index, and negative tests for a conflicting definition, a persistent missing index and duplicate-data failure. Repeat the existing concurrent migration test, then run the surrounding ownership/migration suite and Checkstyle. Submit the fix independently from the conversation-deletion design. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
