Details
-
Test
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
-
Reviewed
Description
The failure to compare timestamp of last applied op with RS start is not a permanent failure, nor does it seem flaky but while running all tests as part of 2.3 RC0 validation, I did encounter the failure:
[ERROR] Tests run: 2, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 29.212 s <<< FAILURE! - in org.apache.hadoop.hbase.replication.TestReplicationStatus[ERROR] Tests run: 2, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 29.212 s <<< FAILURE! - in org.apache.hadoop.hbase.replication.TestReplicationStatus[ERROR] org.apache.hadoop.hbase.replication.TestReplicationStatus.testReplicationStatusSink Time elapsed: 1.016 s <<< FAILURE!java.lang.AssertionError: expected:<1592477704810> but was:<1592477719964> at org.apache.hadoop.hbase.replication.TestReplicationStatus.testReplicationStatusSink(TestReplicationStatus.java:134)
It seems we should allocate separate resources to testReplicationStatusSink: separate class, shutdown peer cluster (if running already), bring it up and check timestamp of lastAppliedOp, then proceed with remaining test i.e timestampsOfLastAppliedOp > timestampStarted within specific period of time soon after inserting some records.
Attachments
Issue Links
- links to