[ovirt-devel] [ ERROR ] schema.sh: Please fix numbering to interval 04010491 to 04010500 and run the upgrade script.

Yedidyah Bar David didi at redhat.com
Thu Nov 24 11:09:13 UTC 2016


On Thu, Nov 24, 2016 at 12:51 PM, Martin Perina <mperina at redhat.com> wrote:
>
>
> On Thu, Nov 24, 2016 at 11:25 AM, Yedidyah Bar David <didi at redhat.com>
> wrote:
>>
>> On Thu, Nov 24, 2016 at 12:21 PM, Martin Perina <mperina at redhat.com>
>> wrote:
>> >
>> >
>> > On Thu, Nov 24, 2016 at 10:37 AM, Martin Perina <mperina at redhat.com>
>> > wrote:
>> >>
>> >>
>> >>
>> >> On Thu, Nov 24, 2016 at 9:31 AM, Eyal Edri <eedri at redhat.com> wrote:
>> >>>
>> >>> Can you paste here example of such job failing now?
>> >>
>> >>
>> >> Sure, please take a look at patch https://gerrit.ovirt.org/66999
>> >>
>> >> I've rebased it yesterday at 23:22 and upgrade from master job failed
>> >> due
>> >> to the error I've described below:
>> >>
>> >>
>> >>
>> >> http://jenkins.ovirt.org/job/ovirt-engine_master_upgrade-from-master_el7_created/9295/
>> >>
>> >> Although at that moment patch
>> >> https://gerrit.ovirt.org/67210 , which fixed faulty revert patch [1],
>> >> was
>> >> already merged, so engine-setup should execute successfully.
>> >>
>> >> I've just scheduled another build on the same patch (there was no
>> >> rebase
>> >> at this time), so you can compare the results.
>> >
>> >
>> > So today's build finished fine, which is strange, because
>> >
>> > patch https://gerrit.ovirt.org/67210 haven't been rebased, so upgrade
>> > from
>> > master should fail same way as yesterday -> so it seems that upgrade
>> > from
>> > master caches somewhere git hash, which is used as a startup for db
>> > upgrade
>> > job and not always using actual latest commit from master
>>
>> AFAIU the upgrade job upgrade from the latest nightly master snapshot
>> build, not from git.
>> So after you merge such a patch and its builds are finished, you have
>> to run the publisher.
>
>
> Well, it that case the logic of "upgrade from master" job is a bit
> misleading, because it's "upgrade from yesterday's master" actually ...

I agree. I guess the intention was to "upgrade from last known good
version of master". Perhaps we should change that. Perhaps something like:

https://gerrit.ovirt.org/67263

>
>>
>>
>> >
>> >>
>> >> Martin
>> >>
>> >>>
>> >>> On Thu, Nov 24, 2016 at 9:42 AM, Martin Perina <mperina at redhat.com>
>> >>> wrote:
>> >>>>
>> >>>> I don't know what exactly upgrade from master job is doing, but
>> >>>> according to error produced by engine-setup it does not take latest
>> >>>> patch
>> >>>> merged to master to prepare initial db on which we try to execute
>> >>>> upgrade.
>> >>>> It take the revert patch, which is faulty and that's why upgrade job
>> >>>> is
>> >>>> failing.
>> >>>>
>> >>>> Martin
>> >>>>
>> >>>> On Thursday, November 24, 2016, Eyal Edri <eedri at redhat.com> wrote:
>> >>>> > What does it mean to clean the jobs?  You mean git history or temp
>> >>>> > files?
>> >>>> >
>> >>>> > On Nov 24, 2016 1:23 AM, "Martin Perina" <mperina at redhat.com>
>> >>>> > wrote:
>> >>>> >>
>> >>>> >> So, I tested current master and also upgrade from 4.0 to master
>> >>>> >> and
>> >>>> >> everything works fine on my local machine.
>> >>>> >>
>> >>>> >> But I've found 2 issues:
>> >>>> >>
>> >>>> >> 1. Jenkins CI for some reason unknown to me doesn't use current
>> >>>> >> master (change-id: I59087ff8902e35b9f7c922eba0a17ba4f0791795) but
>> >>>> >> it uses
>> >>>> >> older commit (most probably revert patch from Eyal [1] with
>> >>>> >> Change-Id:
>> >>>> >> I8c2350aba48619dfa84624b343cfa167d0b6e0ed).as a current master
>> >>>> >> when
>> >>>> >> executing upgrade job from master to patch
>> >>>> >>
>> >>>> >> 2. When we are reverting patches which contain db upgrade scripts
>> >>>> >> which are not last (for example we are dropping db script
>> >>>> >> 04_01_0500 when
>> >>>> >> there's already merged script 04_01_0510) we cannot delete upgrade
>> >>>> >> script
>> >>>> >> otherwise db schema upgrade script fails (revert script [1] is
>> >>>> >> faulty). Only
>> >>>> >> following actions are valid in this case:
>> >>>> >>
>> >>>> >>   a. Comment out code in db upgrade script we want to revert, but
>> >>>> >> leave this script in the repo
>> >>>> >>
>> >>>> >>   b. Remove the db upgrade script and renumber all later db
>> >>>> >> upgrade
>> >>>> >> scripts so there's no hole in db upgrade script numbering
>> >>>> >>
>> >>>> >> Conclusion: current code in master is OK, we need clean up jenkins
>> >>>> >> job to really use latest merged patch on git master branch.
>> >>>> >>
>> >>>> >> Martin
>> >>>> >>
>> >>>> >>
>> >>>> >> [1] https://gerrit.ovirt.org/#/c/67205/
>> >>>> >>
>> >>>> >>
>> >>>> >> On Wed, Nov 23, 2016 at 7:17 PM, Michal Skrivanek
>> >>>> >> <mskrivan at redhat.com> wrote:
>> >>>> >>>
>> >>>> >>>
>> >>>> >>> > On 23 Nov 2016, at 17:44, Martin Sivak <msivak at redhat.com>
>> >>>> >>> > wrote:
>> >>>> >>> >
>> >>>> >>> > Yeah, our patches are still failing as well. It seems the patch
>> >>>> >>> > was
>> >>>> >>> > not enough or there is some other conflicting db script.
>> >>>> >>>
>> >>>> >>> I think the hash changed. IIRC there was some issue in the logic
>> >>>> >>> when the existing script is updated (since it was merged and
>> >>>> >>> reverted and
>> >>>> >>> re-merged with the same number, but it was not the exact same
>> >>>> >>> content)
>> >>>> >>> Martin/Eli to the rescue!
>> >>>> >>>
>> >>>> >>> >
>> >>>> >>> > Martin
>> >>>> >>> >
>> >>>> >>> >> On Wed, Nov 23, 2016 at 2:16 PM, Sandro Bonazzola
>> >>>> >>> >> <sbonazzo at redhat.com> wrote:
>> >>>> >>> >>
>> >>>> >>> >>
>> >>>> >>> >>> On Wed, Nov 23, 2016 at 2:13 PM, Arik Hadas
>> >>>> >>> >>> <ahadas at redhat.com>
>> >>>> >>> >>> wrote:
>> >>>> >>> >>>
>> >>>> >>> >>> it should be fixed now (by
>> >>>> >>> >>> https://gerrit.ovirt.org/#/c/67210/)
>> >>>> >>> >>
>> >>>> >>> >>
>> >>>> >>> >> I rebased https://gerrit.ovirt.org/66999 and still failing
>> >>>> >>> >> there.
>> >>>> >>> >>
>> >>>> >>> >>
>> >>>> >>> >>>
>> >>>> >>> >>> ----- Original Message -----
>> >>>> >>> >>>> engine master CI is failing on $subject, please fix or
>> >>>> >>> >>>> revert
>> >>>> >>> >>>> as soon as
>> >>>> >>> >>>> possible, thanks.
>> >>>> >>> >>>>
>> >>>> >>> >>>> --
>> >>>> >>> >>>> Sandro Bonazzola
>> >>>> >>> >>>> Better technology. Faster innovation. Powered by community
>> >>>> >>> >>>> collaboration.
>> >>>> >>> >>>> See how it works at redhat.com
>> >>>> >>> >>>>
>> >>>> >>> >>>> _______________________________________________
>> >>>> >>> >>>> Devel mailing list
>> >>>> >>> >>>> Devel at ovirt.org
>> >>>> >>> >>>> http://lists.ovirt.org/mailman/listinfo/devel
>> >>>> >>> >>
>> >>>> >>> >>
>> >>>> >>> >>
>> >>>> >>> >>
>> >>>> >>> >> --
>> >>>> >>> >> Sandro Bonazzola
>> >>>> >>> >> Better technology. Faster innovation. Powered by community
>> >>>> >>> >> collaboration.
>> >>>> >>> >> See how it works at redhat.com
>> >>>> >>> >>
>> >>>> >>> >> _______________________________________________
>> >>>> >>> >> Devel mailing list
>> >>>> >>> >> Devel at ovirt.org
>> >>>> >>> >> http://lists.ovirt.org/mailman/listinfo/devel
>> >>>> >>> > _______________________________________________
>> >>>> >>> > Devel mailing list
>> >>>> >>> > Devel at ovirt.org
>> >>>> >>> > http://lists.ovirt.org/mailman/listinfo/devel
>> >>>> >>> >
>> >>>> >>> >
>> >>>> >>
>> >>>> >
>> >>>
>> >>>
>> >>>
>> >>>
>> >>> --
>> >>> Eyal Edri
>> >>> Associate Manager
>> >>> RHV DevOps
>> >>> EMEA ENG Virtualization R&D
>> >>> Red Hat Israel
>> >>>
>> >>> phone: +972-9-7692018
>> >>> irc: eedri (on #tlv #rhev-dev #rhev-integ)
>> >>
>> >>
>> >
>> >
>> > _______________________________________________
>> > Devel mailing list
>> > Devel at ovirt.org
>> > http://lists.ovirt.org/mailman/listinfo/devel
>>
>>
>>
>> --
>> Didi
>
>



-- 
Didi



More information about the Infra mailing list