[
https://issues.apache.org/jira/browse/HBASE-18146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16033616#comment-16033616
]
Andrew Purtell commented on HBASE-18146:
----------------------------------------
Looking at the code, IntegrationTestAcidGuarantees wants to apply special
configuration to the minicluster and then run three relatively short canned
scenarios.
It's good to know someone is using it on real clusters and with chaos. Do you
think it sufficient or could there be improvement? I'm pretty sure there is
room for improvement for testing against real clusters. One thing I was
thinking regards a new IT is a new monkey policy that flushes and compacts very
frequently.
> Long running integration test similar to TestAcidGuarantees
> -----------------------------------------------------------
>
> Key: HBASE-18146
> URL: https://issues.apache.org/jira/browse/HBASE-18146
> Project: HBase
> Issue Type: Test
> Components: integration tests
> Reporter: Andrew Purtell
>
> TestAcidGuarantees and IntegrationTestAcidGuarantees both really only work
> with minicluster based testing and do not run for a long duration. Consider a
> new integration test that makes similar atomicity checks while running for,
> potentially, a very long time, determined by test parameters supplied on the
> command line (perhaps as property definitions). The new integration test
> should expect to run against a distributed cluster, support specification of
> desired monkey policy, and not require any special non-default site
> configuration settings.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
