CRZbulabula opened a new pull request, #18699:
URL: https://github.com/apache/iotdb/pull/18699
## Description
After a ConfigNode leader switch, a region migration procedure can resume
before the Region leader cache is rebuilt. Both replica selection and leader
transfer currently unbox a missing cache entry, causing a
`NullPointerException` that terminates `RemoveRegionPeerProcedure` before
removal completes.
Treat a missing entry as the existing unknown-leader value (`-1`) in both
lookups. This lets replica selection choose an eligible peer and lets Ratis
leader transfer proceed when the current leader is unknown. Cached leaders
retain their existing preference and transfer behavior.
Add regression tests for DataRegions and SchemaRegions covering missing and
explicitly unknown leaders, known-leader preference, excluded and unrelated
nodes, successful leader transfer and cache updates, and a cache entry
disappearing between the two lookups.
## Validation
- Before the fix, the new regression suite reproduced eight
`NullPointerException`s across both Region types, including failures at each of
the two leader lookups.
- After the fix, all 14 new cases and 12 existing related cases pass. The
full reactor build succeeds with:
```sh
mvn -B test \
-Dtest=RegionMaintainHandlerTest,RegionMaintainHandlerConsensusPipeTest,RemoveRegionPeerProcedureTest,RegionMigrateProcedureTest
\
-Dsurefire.failIfNoSpecifiedTests=false -DfailIfNoTests=false
```
- `mvn -B spotless:apply -pl iotdb-core/confignode` passes.
- `mvn -B test-compile -DskipTests -P with-zh-locale` passes for the full
reactor.
Cluster failover integration testing was not run.
This PR has:
- [x] been self-reviewed.
- [x] added unit tests to cover the affected code paths.
## Key changed/added classes
- `RegionMaintainHandler`
- `RegionMaintainHandlerTest`
--
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]