[ovirt-devel] [ OST Failure Report ] [ ovirt-engine-4.1 ] [ 02.03.2017 ] [004_basic_sanity]

[Eyal Edri]

On Sun, Mar 19, 2017 at 10:33 AM, Eyal Edri <eedri at redhat.com> wrote:

>
>
> On Sun, Mar 19, 2017 at 10:30 AM, Oved Ourfali <oourfali at redhat.com>
> wrote:
>
>> As Piotr mentioned, the job is using 1.3.8 however 1.3.9 was released.....
>>
>
> Thanks,
> We will look into the logs and find out what caused it to use 1.3.8 and
> report here on the results, can't be sure yet if its a bug in the job or
> anything else.
>

Let me explain what happened:

1. check-patch jobs used to run on latest 'tested' repo, but that repo is
very big as it contains history of all the builds done for a release [1]
2. recently we found a bug that caused OST to fail due to out of memory
because it tried to sync the entire tested repo into memory ( that's how
OST works ), so the tested repo was removed from check-patch.
3. This means that now check-patch is only using the repos in the reposync
file, which for 4.1 is latest released.
4. vdsm-jsonrpc-java 1.3.9 wasn't released in any official oVirt 4.1
release, this is why the job used 1.3.8 which is the latest released.
5. We also don't want to replace the default repos for OST on 4.1 to
something other than released, because the basic suite per release should
always check against official release [2]
6. As a side note - manual jobs also support running on latest tested, so
this issue is solely restricted to 'check-patch' jobs.

We do however believe that running check-patch on latest code is important,
so we'll look into providing an alternate solution that won't break any of
the above restrictions,
Suggestions are welcome of course.



[1] It is done so anyone that wants to check oVirt  can do it w/o  worrying
that a pkg will be removed during testing ( something that was happening
when used experimental lastest.tested repo )
[2] As opposed to experimental flows which DO check against latest code,
which is why 4.1 experimental didn't fail on this issue, as it used 1.3.9



>
>
>>
>> On Sun, Mar 19, 2017 at 10:27 AM, Barak Korren <bkorren at redhat.com>
>> wrote:
>>
>>>
>>>
>>> בתאריך 17 במרץ 2017 02:32 PM,‏ "Piotr Kliczewski" <
>>> piotr.kliczewski at gmail.com> כתב:
>>>
>>>
>>> The change was merged 2 weeks ago. How can we make sure not to waste
>>> people time next time to analyze the issue?
>>>
>>>
>>> "Merged" is hardly "released".
>>>
>>> As long as a fix does not make it to a release, users can still come
>>> across the issue, and so can OST. I can't think of a quick way to avoid
>>> that, but I'm open to suggestions....
>>>
>>> WRT getting more frequent oVirt releases, I'm hardly in a position to
>>> give input about that.
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> 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 <+972%209-769-2018>
> irc: eedri (on #tlv #rhev-dev #rhev-integ)
>



-- 
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)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/devel/attachments/20170319/06477e25/attachment.html>