You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@phoenix.apache.org by "Chinmay Kulkarni (Jira)" <ji...@apache.org> on 2020/11/20 03:34:00 UTC
[jira] [Comment Edited] (PHOENIX-6230) IT suite hangs on
ViewConcurrencyAndFailureIT
[ https://issues.apache.org/jira/browse/PHOENIX-6230?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17235855#comment-17235855 ]
Chinmay Kulkarni edited comment on PHOENIX-6230 at 11/20/20, 3:33 AM:
----------------------------------------------------------------------
[~vjasani] [~stoty] I see the problem. The patch in master for 6199 missed [this part|https://github.com/apache/phoenix/blob/d0c3cafc13874093a513b54b4894897aabc2d0af/phoenix-core/src/it/java/org/apache/phoenix/end2end/ViewIT.java#L163-L166] which is only required in master branch. Because of this, the TestMetaDataRegionObserver code doesn't get invoked at all since [this condition|https://github.com/apache/phoenix/blob/d0c3cafc13874093a513b54b4894897aabc2d0af/phoenix-core/src/main/java/org/apache/phoenix/coprocessor/PhoenixMetaDataCoprocessorHost.java#L85] inside PhoenixMetaDataCoprocessorHost becomes false.
Looks like 4.x and master are diverging more and more with time and so it isn't safe to just apply a successful patch of 4.x directly on master (of course after resolving conflicts, API changes, etc.) since such subtle differences can be left out.
So far I've seen contributors put up a PR/patch for just one of the 2 branches and then just modify their patch for the other branch (in case of API diffs), but looks like we should now enforce 2 separate PRs for each contribution (one for master and one for 4.x) to avoid more such issues in the future?
I also found another smaller issue which I will fix with my patch and will put up a pr for this and a separate one for the 4.x changes.
was (Author: ckulkarni):
[~vjasani] [~stoty] I see the problem. The patch in master for 6199 missed [this part|https://github.com/apache/phoenix/blob/d0c3cafc13874093a513b54b4894897aabc2d0af/phoenix-core/src/it/java/org/apache/phoenix/end2end/ViewIT.java#L163-L166] which is only required in master branch. Because of this, the TestMetaDataRegionObserver code doesn't get invoked at all since [this condition|https://github.com/apache/phoenix/blob/d0c3cafc13874093a513b54b4894897aabc2d0af/phoenix-core/src/main/java/org/apache/phoenix/coprocessor/PhoenixMetaDataCoprocessorHost.java#L85] inside PhoenixMetaDataCoprocessorHost becomes false.
Looks like 4.x and master are diverging more and more with time and so it isn't safe to just apply a successful patch of 4.x directly on master (of course after resolving conflicts, API changes, etc.) since such subtle differences can be left out. So far I've seen committers put up a PR/patch for just one of the 2 branches and then just modify their patch for the other branch (in case of API diffs), but looks like we should now enforce 2 separate PRs for each contribution (one for master and one for 4.x) to avoid more such issues in the future?
I also found another smaller issue which I will fix with my patch and will put up a pr for this and a separate one for the 4.x changes.
> IT suite hangs on ViewConcurrencyAndFailureIT
> ---------------------------------------------
>
> Key: PHOENIX-6230
> URL: https://issues.apache.org/jira/browse/PHOENIX-6230
> Project: Phoenix
> Issue Type: Bug
> Components: core
> Affects Versions: 5.0.0
> Reporter: Istvan Toth
> Assignee: Chinmay Kulkarni
> Priority: Critical
> Fix For: 5.1.0
>
>
> The ASF Jenkins postcommit job timed out 5 times out of the last 6 runs on the master branch.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)