[JIRA] (OVIRT-1500) CI on ovirt-engine patch succeeded, but didn't mark CI=+1 on gerrit

[Barak Korren (oVirt JIRA)]

    [ https://ovirt-jira.atlassian.net/browse/OVIRT-1500?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=33144#comment-33144 ] 

Barak Korren commented on OVIRT-1500:
-------------------------------------

[~eedri] Its the gerrit plugin, but it simple runs a gerrit command on ssh. You can try it yourself, send a command to update the vote for a patchset that is not the most recent one, Gerrit will add a comment that says you updated the vote but not update it in practice because it only keeps votes for the most recent  patcheset.

What we have here is a race condition - A new patchset was added before jenkins managed to add its vote, and because the new patchset was a trivial rebase no new jenkins job was invoked.

A hook to copy the votes will not help here because they were not registered by Gerrit in the 1st place. The only solution I can think of ATM is to make Jenkins run jobs on trivial rebase too.

> CI on ovirt-engine patch succeeded, but didn't mark CI=+1 on gerrit
> -------------------------------------------------------------------
>
>                 Key: OVIRT-1500
>                 URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1500
>             Project: oVirt - virtualization made easy
>          Issue Type: Bug
>          Components: Gerrit/git, oVirt CI
>            Reporter: Allon Mureinik
>            Assignee: infra
>
> In my recent engine patch https://gerrit.ovirt.org/#/c/78911 the CI passed, but was not marked as so in gerrit.
> Looking through gerrit's history, here's what I *think* is going on:
> 08:41 - patchset #1 is submitted
> 08:41 - CI (check-patch) starts for both EL7 and fc25
> _At this point I noticed a typo in the commit method. I fixed it and posted patchset #2_
> 08:42 - patchset #2 is submitted
> _CI was never triggered for patchset #2. Is this because the only difference between it and patchset #1 is the commit message?_ 
> 09:15 - CI for patchset #1 completes, and marks *patchset #1* with CI=+1. This is not inherited to patchset #2 for some reason.
> Expected behavior:
> If CI for patchset #1 passed and the only difference is the commit message (not the parent commit and not the content), I'd expect the CI score to be inherited.



--
This message was sent by Atlassian JIRA
(v1000.1092.1#100053)