[lago-devel] Weekly bug status for ovirt-system-tests (using Lago)

[Yaniv Kaul]

On Sun, Apr 3, 2016 at 11:47 AM, Eyal Edri <eedri at redhat.com> wrote:

> A weekly reminder on open bugs & patches to keep track on blockers for
> oVirt testing:
>
>
> *Open bugs:*
> https://goo.gl/eUaPUY
>
>
> *Blocking bugs in potential:*
>
>    - Bug 1311121 - oVirt system tests for 3.6 fail to deactivate domain
>    on the first try - suspect race between OVF update and SPM stop (engine bug)
>    - Patch provided, pending verification/testing on the tests
>
>
Looks like https://gerrit.ovirt.org/#/c/55514/ [Basic quota tests] - which
has nothing to do with storage, can be a great workaround for the above.
Since essentially the above is a race, just doing something that takes a
bit of time will allow the OVF update to be completed successfully.
Playing with quota allows that.
In addition, a simple switch between the order of hot-add disk and hot-add
NIC in 004 tests will overcome the 2nd place this might happen.


>
> *Open patches:*
> https://gerrit.ovirt.org/#/q/project:ovirt-system-tests+status:open
>
>
>    - WIP:Fix the master suite runs
>    - Improvements to storage domains add tests
>    - Merged storage servers into one
>    - Basic quote tests
>
>
> If you have any updates on the bugs/patches please reply with more info.
>

The move to EL7 has brought in several issues:
1. For some reason, there's now a need to sync tens of more packages. Not
sure why. I'll send a patch later this week to do that. With them, host
installation time is slashed almost back to normal.
2. 'Almost back to normal' - as there's another regression (for which I
have a workaround as well) that needs to be handled too, to take host
installation down to ~2 minutes.
3. Hosts stuck on installation[1] - I believe this is a network bug.

In addition:
4. I think that writing logs into test_logs/... is a bit of a regression -
it doesn't allow to easily run multiple tests, as I've used to. Offline
discussion with David turned out that there's some additional work to do to
allow it - we need to focus on it (after uniting the storage servers, etc.)
5.  I have several changes to the storage server deployment scripts - small
changes, but I believe they are beneficial (move everything to XFS, fewer
LUNs, disable firewall on them, etc.)
6. Seems like I've almost managed to install Gluster in a hyper-converged
setup with gdeploy[2]. I'll try to push this too this week. It's a bit
ugly, but works.
7. I'm looking into collect artifacts. Just realized it freezes the OS,
which isn't optimal. I've tried with scp only - and it looks good and fast.

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1322257
[2] https://github.com/gluster/gdeploy

>
>
> --
> Eyal Edri
> Associate Manager
> RHEV DevOps
> EMEA ENG Virtualization R&D
> Red Hat Israel
>
> phone: +972-9-7692018
> irc: eedri (on #tlv #rhev-dev #rhev-integ)
>
> _______________________________________________
> lago-devel mailing list
> lago-devel at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/lago-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/lago-devel/attachments/20160403/a882b7f4/attachment-0001.html>