[
https://issues.apache.org/jira/browse/PHOENIX-4143?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16149837#comment-16149837
]
James Taylor commented on PHOENIX-4143:
---------------------------------------
+1 with one minor nit: set an upper limit on how long you'll continue looping
and fail the test if that is reached.
> ConcurrentMutationsIT flaps
> ---------------------------
>
> Key: PHOENIX-4143
> URL: https://issues.apache.org/jira/browse/PHOENIX-4143
> Project: Phoenix
> Issue Type: Bug
> Reporter: Samarth Jain
> Assignee: Samarth Jain
> Attachments: PHOENIX-4143.patch
>
>
> I have seen sometimes that the test flaps with a RegionOpeningException
> especially on master and 1.x branches. I believe this is likely because we
> are adding a co-processor by doing an admin.modify() call which is
> asynchronous. So it's possible that the region would likely be still opening
> even though the call returned. A potential fix is to poll for the updated
> table descriptor. Not sure if it will always prevent test from flapping
> (since the problem is similar to what Rajesh is solving in PHOENIX-3496) but
> it should reduce the chances.
> FYI, [~jamestaylor]
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
