[HBASE-23880] [Flakey Test] TestRegionReplicaFailover fails frequently

• Type: Bug
• Status: Resolved
• Priority: Trivial
• Resolution: Incomplete
• Affects Version/s: 3.0.0-alpha-1, 2.3.0
• Component/s: flakies
• Assignee: Unassigned
• Reporter: Michael Stack
• Created: 21/Feb/20 06:12
• Updated: 21/Feb/20 22:21
• Resolved: 21/Feb/20 17:40

Description

Fails > 50% of the time in the GCE runs since we upped their ferocity a few days ago.

Looking, the complaints are weirdly often OOME, can't create thread. Looking, it at least does start of a RS at end of a test when it kills one EVEN THOUGH the @before starts a whole new cluster. This improves general resource usage going by the resource dumps before and after but still loads of leaking.

Cut down on the amount of work it does too. Let me push small patch and see if helps.