[
https://issues.apache.org/jira/browse/PHOENIX-3036?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Samarth Jain updated PHOENIX-3036:
----------------------------------
Attachment: PHOENIX-3036_addendum.patch
Addendum patch to address verification failures we were running into.
It took me a while to figure out what exactly was going on here since failsafe
wasn't reporting the "real" error behind why the test forks were unsuccessful.
After unsuccessfully hacking failsafe (including hacking the streams through
which parent and child processes communicate), I had to resort to the old
fashion way of running all the tests in a junit test suite, bumping up the
memory and pruning the logs to see why possibly are forked JVMs
hanging/crashing. Turns out IndexIT#testMultipleUpdatesAcrossRegions had
MAX_FILESIZE and MEMSTORE_SIZE set to 1! The test really doesn't need it since
it already is splitting the table at split points. I also noticed that we are
running into constraint for old-gen JVM heap space. This is likely because we
are creating a lot many tables now and not dropping them. I have tweaked the
memory settings to address that. Also, got rid of the annoying log messages
that cribbed (rightfully) about Phoenix using a deprecated version of client
scanner timeout config.
> Modify phoenix IT tests to extend BaseHBaseManagedTimeTableReuseIT
> ------------------------------------------------------------------
>
> Key: PHOENIX-3036
> URL: https://issues.apache.org/jira/browse/PHOENIX-3036
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Samarth Jain
> Assignee: prakul agarwal
> Fix For: 4.9.0
>
> Attachments: PHOENIX-3036.patch, PHOENIX-3036_addendum.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
