[
https://issues.apache.org/jira/browse/PHOENIX-6850?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Geoffrey Jacoby updated PHOENIX-6850:
-------------------------------------
Labels: beginner starter (was: )
> AlterTableWithViewsIT CreateView Props Test Flaps
> -------------------------------------------------
>
> Key: PHOENIX-6850
> URL: https://issues.apache.org/jira/browse/PHOENIX-6850
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 5.2.0, 5.1.3
> Reporter: Geoffrey Jacoby
> Priority: Major
> Labels: beginner, starter
>
> When running IT tests on the 5.1.3 RC1, and on the master (5.2) HEAD, I get
> flappy behavior on
> AlterTestsWithViewsIT.testCreateViewWithPropsMaintainsOwnProps. When a
> particular param set is run standalone, it seems to consistently pass.
> However, when run in concert with different param iterations, it sometimes
> generates an NPE on
> {code:java}
> assertFalse(viewTable1.useStatsForParallelization());
> {code}
> This is because viewTable1 had previously been unset for
> useStatsForParallelization, so it returns null if it doesn't pick up the
> change to the base table properly.
> This seems to be a caching problem -- populating viewTable1 and viewTable2
> from a call to PhoenixRuntime.getTableNoCache seems to fix it.
> However, since the test updates the base table from a global connection, and
> then tries to access views on that table from a separate tenant connection,
> it's not obvious to me that the cache for the tenant connection _should_ be
> expired in this situation, so I'm not sure the caching behavior counts as a
> bug itself.
> Interestingly though, 5.1.2 doesn't seem to have this issue.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
