Storage functional tests CI job

[Vered Volansky]

Hi David,

----- Original Message -----
> From: "David Caro" <dcaroest at redhat.com>
> To: "Vered Volansky" <vered at redhat.com>, "infra" <infra at ovirt.org>
> Sent: Wednesday, May 28, 2014 2:21:45 PM
> Subject: Re: Storage functional tests CI job
> 
> On 05/26/2014 02:58 PM, Vered Volansky wrote:
> > Hi David,
> > 
> > After talking with Eyal and Ohad we understand our needs comply with a
> > matrix job.
> Upstream or downstream?
upstream.
> 
> Which project/projects? Should it build engine/vdsm/both.
Currently the working job is vdsm_master_storage_functional_tests_localfs_gerrit, but we're working vdsm_master_storage_functional_tests_nfs_gerrit as well, and a new job Yoav requested based on 
vdsm_master_storage_functional_tests_localfs_gerrit . So instead all of the above and more we were thinking about a matrix job based on vdsm_master_storage_functional_tests_localfs_gerrit.
We're only building vdsm.
> 
> Should it use a nightly repo or something?
No, just building vdsm.
> 
> Should it be triggered by ovirt-engine patches? vdsm patches? both? Timed?
> manually run only?
It should run on every patchset, as does vdsm_master_storage_functional_tests_localfs_gerrit.
> 
> If triggered by gerrit, should it send any feedback to the patch? Should it
> block the patch if failed?
Block, again as already configured in vdsm_master_storage_functional_tests_localfs_gerrit.
> 
> Which setup is required for the tests to start?
> (ex. build vdsm, install packages, services up, network interfaces...)
build vdsm, some "cells" in the matrix should start services (rpcbind, nfs) we then stop at cleanup, some don't.
Some might need network interfaces in the future, I'm not currently working on such a test.
> 
> What needs to be cleaned up after?
Right now - only stopping the above services, removing vdsm installation.
The current job was configured based on some network job, since we didn't exactly what we should be doing, so I'm hopfull we don't still have garbage there.
> 
> Can it run in parallel with itself? And with other jobs?
No.
> 
> Does it require a specific hardware/distribution?
fedora 19/20, specifically 20.
> 
> What should be saved from the tests (any specific logs? Just the output?)
vdsm and supervdsm logs. already set to be cleaned at beginning and saved. 
> 
> > Our inputs (currently passed by env variables to run_tests.sh) are
> > generally the cross product of version and storage types:
> > version=[0,1,2,3], storage_type=[localfs,nfs,posix,iscsi,fc].
> Do you have an example of one of the combinations run script?
> (ex. ./run_tests.sh --storage-type localfs --version 1)
currently we're running it based on env vars:
sudo NOSE_WITH_XUNIT=1 \
     VDSM_TEST_STORAGE_TYPES='localfs' \
     VDSM_TEST_STORAGE_VERSIONS='0 3' \
     ./run_tests.sh  functional/storageTests.py
> 
> > Can you please set a time for us to set this up together?
> Sure, but have the answers for the previous questions beforehand, so we can
> start working right away, I don't know anything of the above, so I can't help
> defining them.
All is basically already answered in the already existing job:
vdsm_master_storage_functional_tests_nfs_gerrit .
We need the matrix job to be based on this condiguration.
> 
> > We'll need Yoav Kleinberger to have permissions for this job as well.
> > Dependent on how long it takes to set this up, this job might be enough,
> > with no need to the job Yoav asked in the mailing list.
> > 
> > Thanks,
> > Vered
> > 
> 
> 
> --
> David Caro
> 
> Red Hat S.L.
> Continuous Integration Engineer - EMEA ENG Virtualization R&D
> 
> Email: dcaro at redhat.com
> Web: www.redhat.com
> RHT Global #: 82-62605
> 
> 
> _______________________________________________
> Infra mailing list
> Infra at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/infra
>