[
https://issues.apache.org/jira/browse/HIVE-5695?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13809225#comment-13809225
]
Brock Noland commented on HIVE-5695:
------------------------------------
1) Propagating a shutdown flag seems appropriate
2) We can store an attachment on the JIRA and then only reset if the to be
tested attachment is not stored on the jira
3) We just need to add a more aggressive retry handler for these ops
> PTest2 fix shutdown, duplicate runs, and add client retry
> ---------------------------------------------------------
>
> Key: HIVE-5695
> URL: https://issues.apache.org/jira/browse/HIVE-5695
> Project: Hive
> Issue Type: Bug
> Reporter: Brock Noland
> Assignee: Brock Noland
>
> There are three issues with the PTest2 framework at present:
> 1) When a test crashes it doesn't shutdown the underlying executors quickly
> 2) If a jira has two patches uploaded say 20 minutes a part that jira will
> have two pre-commit runs executed
> 3) The client doesn't aggressively retry impotent operations leading to the
> client (i.e. jenkins) finishing before the test actually does
--
This message was sent by Atlassian JIRA
(v6.1#6144)
