apache
PHOENIX-6714 Return update status from Conditional Upserts
@jinggou I think there seems to be some formatting changes to IndexRegionObserver.java which has made the diff huge. Can we revert and just keep what's relevant.
@tkhurana @virajjasani Please review my changes
- reverted the formatting changes to IndexRegionObserver.java and keep what's relevant (resolved merge conflicts)
- allocated results array only when autocommit is set and batch size is 1
- added tests for batched upsert
HBase 2.6 support is likely going to land very soon https://github.com/apache/phoenix/pull/1793
Also, 2.4.18 is already released: https://github.com/apache/hbase/releases/tag/rel%2F2.4.18 2.5.9 is about to be released soon.
@jinggou could you please remove @ignore for tests and apply logic similar to https://github.com/apache/phoenix/blob/master/phoenix-core/src/it/java/org/apache/phoenix/end2end/index/ReplicationWithWALAnnotationIT.java#L173-L201 ?
For the tests, we will need to keep min versions: 2.6.0, 2.5.9 and 2.4.18. For versions below them, test should be ignored automatically as per the logic.
All the latest builds are timing out, aborting or getting much longer to complete. The cause is also not clear and tests are passing locally. We should try breaking OnDuplicateKeyIT into two classes: move half of the tests to new class OnDuplicateKeyIT2.
It's still not going well, the latest build is running for almost 6 hr, it will either get aborted or timed out: https://ci-hadoop.apache.org/job/Phoenix/job/Phoenix-PreCommit-GitHub-PR/job/PR-1884/21/
can see that the test was running when the OutOfMemoryError was thrown from logs: https://ci-hadoop.apache.org/job/Phoenix/job/Phoenix-PreCommit-GitHub-PR/job/PR-1884/21/artifact/yetus-general-check/output/patch-unit-root.txt
[INFO] Running org.apache.phoenix.end2end.OnDuplicateKeyIT