Re: [Gluster-infra] All rpm jobs are now in jenkins job builder

2016-06-29 Thread Sankarshan Mukhopadhyay 
The detail in this response is much appreciated. Thank you.

On Wed, Jun 29, 2016 at 5:25 PM, Kaleb S. KEITHLEY  wrote:
> For Fedora/RHEL/CentOS rpms:
>
>  + People file bugs, e.g. against Fedora/glusterfs or
> GlusterFS/packaging, and we fix them.
>
>  + Some of us  occasionally do package reviews for new packages and/or
> closely follow the Fedora packaging guidelines and when we see something
> that ought to be done in the glusterfs packaging we fix it.
>
>  + Every once in a while I run rpmlint and address the things it finds.
>
>  + And of course we get feedback from downstream packaging.
>
>  + Finally, I  keep the Fedora dist-git .spec and our upstream .spec in
> sync.
>
> For SuSE RPMs I use a .spec file based on the one that SuSE uses/used
> for their distribution's bundled packages.
>
> For Debian/Ubuntu debs I use packaging bits provided by Louis Zuckerman
> (irc nick: semiosis) that he developed with, I believe, the help of
> Patrick Matthaei, the Debian packager who builds Debian's bundled
> packages. Resyncing with Patrick's packaging bits is on my list of
> things to do in my copious spare time. In the mean time people
> occasionally report issues with the debs and I fix them.




-- 
sankarshan mukhopadhyay

___
Gluster-infra mailing list
Gluster-infra@gluster.org
http://www.gluster.org/mailman/listinfo/gluster-infra

Re: [Gluster-infra] All rpm jobs are now in jenkins job builder

2016-06-29 Thread Kaleb S. KEITHLEY 
On 06/28/2016 08:07 PM, Sankarshan Mukhopadhyay wrote:
> 
> I have a follow-up question on the production of these artifacts -
> when do we check whether the RPMs or, the images produced are sane?
> For example, that the RPMs are packaged well and as per specifications
> ...
> 

For Fedora/RHEL/CentOS rpms:

 + People file bugs, e.g. against Fedora/glusterfs or
GlusterFS/packaging, and we fix them.

 + Some of us  occasionally do package reviews for new packages and/or
closely follow the Fedora packaging guidelines and when we see something
that ought to be done in the glusterfs packaging we fix it.

 + Every once in a while I run rpmlint and address the things it finds.

 + And of course we get feedback from downstream packaging.

 + Finally, I  keep the Fedora dist-git .spec and our upstream .spec in
sync.

For SuSE RPMs I use a .spec file based on the one that SuSE uses/used
for their distribution's bundled packages.

For Debian/Ubuntu debs I use packaging bits provided by Louis Zuckerman
(irc nick: semiosis) that he developed with, I believe, the help of
Patrick Matthaei, the Debian packager who builds Debian's bundled
packages. Resyncing with Patrick's packaging bits is on my list of
things to do in my copious spare time. In the mean time people
occasionally report issues with the debs and I fix them.

-- 

Kaleb
___
Gluster-infra mailing list
Gluster-infra@gluster.org
http://www.gluster.org/mailman/listinfo/gluster-infra

[Gluster-infra] Adding more jobs to test Gluster regularly

2016-06-29 Thread Niels de Vos 
[and yes, NOW with gluster-devel-on CC, sorry for the duplicate]
[changed subject and added gluster-devel on CC]

On Wed, Jun 29, 2016 at 05:37:17AM +0530, Sankarshan Mukhopadhyay wrote:
> On Tue, Jun 28, 2016 at 9:45 PM, Niels de Vos  wrote:
> > Coincidentally I've asked Humble about the option to provide a container
> > (and maybe VM) image through the CentOS Storage SIG. Just as with the
> > packages, we should try to utilize the integration with different
> > distributions.
> 
> I agree. Containers (and even VM) images are build-time artifacts
> which we should produce and make available in a regular manner.
> 
> I have a follow-up question on the production of these artifacts -
> when do we check whether the RPMs or, the images produced are sane?
> For example, that the RPMs are packaged well and as per specifications
> ...

Well, the RPMs can have different guidelines depending on the
distribution they are made for. So the official RPMs are packaged in
Fedora [1], CentOS Storage SIG [2] and other distribution specific
repositories. The distributions are responsible for verification of the
packages they ship.

But, we could do something like this for the included .spec. The nightly
builds [3] can be consumed by other tests. It is possible to download
the RPMs and run "rpmlint" and other verification tools on them. It is
relatively straight forward to write a script that we can include as a
job in the CentOS CI [4]. The gluster_libgfapi-python script [5] can be
taken as an example.

Anyone is more than welcome to send me scripts that I can wrap in the
needed CentOS CI Jenkins environment. Depending on the time that the
tests need, we can run them every night, week, or whatever. If needed we
can run them upon each patch submission, but we'll probably need to
request a higher resource limit in that case.

I guess you could request an RPM-verification test [6], but please add
some details on what tools you would like to see used. At the moment
there are very few contributors helping out with the running of
automated tests, so do not have too high expectations on when someone
gets around to write+test a script from scratch.

Cheers,
Niels


1. http://pkgs.fedoraproject.org/cgit/glusterfs.git/
2. https://github.com/CentOS-Storage-SIG/glusterfs
3. http://artifacts.ci.centos.org/gluster/nightly/
4. https://ci.centos.org/view/Gluster/
5. 
https://github.com/gluster/glusterfs-patch-acceptance-tests/blob/master/centos-ci/libgfapi-python/run-test.sh
6. https://github.com/gluster/glusterfs-patch-acceptance-tests/issues/new


signature.asc
Description: PGP signature
___
Gluster-infra mailing list
Gluster-infra@gluster.org
http://www.gluster.org/mailman/listinfo/gluster-infra