[
https://issues.apache.org/jira/browse/PHOENIX-2575?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15178162#comment-15178162
]
Rajeshbabu Chintaguntla commented on PHOENIX-2575:
--------------------------------------------------
bq. Did you know this when HBASE-14643 was patched?
I didn't know this James. Just while analyzing the logs of hang issue found
this issue.
bq. So in existing Phoenix LocalIndex code this call itself was not done?
No Ram we are not using the the call in local index.
bq. Now calling super.getLastKey() - does not impact the local index behaviour?
It should not have any impact because we are calling getLastKey() only when mid
key is not null. Here we are retuning mid key as null.
bq. Because previouly u did not want that to be called at all but now you r
using that result.
It's not called in phoenix but after HBASE-14643 it's called just after
creating the reader.
bq. We'll need to adjust the code that determines when local indexing is
supported in ConnectionQueryServicesImpl and update the website page on
secondary indexes with this information too:
Even after adding we need to ask the users to delete if there are any local
indexes because this will effect internally in server during compaction when
there are splits.
> Prevent LocalIndexIT from hanging indefinitely
> ----------------------------------------------
>
> Key: PHOENIX-2575
> URL: https://issues.apache.org/jira/browse/PHOENIX-2575
> Project: Phoenix
> Issue Type: Bug
> Reporter: Samarth Jain
> Assignee: Rajeshbabu Chintaguntla
> Fix For: 4.8.0
>
> Attachments: PHOENIX-2575.patch, PHOENIX-2575.patch
>
>
> On one of the test runs on 4.x-HBase-1.0, I see that the LocalIndexIT class
> didn't complete. Run -
> https://builds.apache.org/job/Phoenix-4.x-HBase-1.0/300/console. This is an
> attempt to prevent the indefinite hang by adding a timeout to count down
> latches.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
