[
https://issues.apache.org/jira/browse/HBASE-25690?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17391807#comment-17391807
]
Nick Dimiduk commented on HBASE-25690:
--------------------------------------
One criteria that comes to my mind is use of a data correctness test (i.e.,
ITBLL) in the face of Chaos. We've never had resources in Apache to run such a
test on a regular cadence, so to use this as a criteria, we'd need to rely on
community-reported experiences. Perhaps we could include some criteria for
launch parameters and successful completion for such a job as part of our
"stable" definition.
> Specify and document the criteria for moving the 'stable' pointer
> -----------------------------------------------------------------
>
> Key: HBASE-25690
> URL: https://issues.apache.org/jira/browse/HBASE-25690
> Project: HBase
> Issue Type: Task
> Components: documentation
> Reporter: Andrew Kyle Purtell
> Priority: Major
>
> We can consider reported experience with new code lines released ahead of
> where the stable pointer currently points, when and if it happens but this
> should not be a formal criteria because it has nothing directly to do with
> our PMC or project. We need a criteria for deciding when new code lines are
> sufficiently stable as to warrant moving the 'stable' pointer that we as
> project and PMC can achieve and implement effectively.
> IMHO "one of our project devs has it running" does not meet that requirement,
> because this depends on third party organizations (a dev's employer, and
> such) and idiosyncratic criteria.
> Collect and document requirements, test scenarios, and success (and failure)
> criteria.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
