I have edited /etc/mrepo.conf and removed it from there as well.
Currently mrepo for external repositories is commented out from cron, so
I believe it is not running at all.
Also I moved postgres backups from /mnt/koji to /mnt/tmp and created a
symlink which gave us 400 MB of free space. Fair enough until jblazek
returns back from his PTO.
Postgres was restarted and koji is operational.
And the most important thing - I am going to call it a day.
Good luck everyone
In fact I have still some work to do on the discovery image…
koji went down this night again. The /mnt/koji folder was filled which
brought our pgsql database down.
I moved the /mnt/koji/pgsql directory to /mnt/tmp where we have plenty
of space now and changed the /var/lib/pgsql symlink. This should prevent
the situation when full disks brings it down again.
I am leaving /mnt/koji/pgsql-ORIG.tar.gz tarball with the old content
just for case something goes wrong. The database files are 1 GB in size
fyi.
Decommissioning our koji is our priority in the upcoming weeks. We need
to look at Copr and evaluate if it fits our needs. We are looking for
public VM hosting with enough disk space (1 TB data volume) if Copr does
not work for us - if you can help drop us a line.
FYI we only dropped Fedora i386 build roots, not the EPELs.
LZ
···
On Mon, Sep 01, 2014 at 08:47:19PM +0200, Lukas Zapletal wrote:
> I see few builds failed since then, mostly jenkins.pulp user, only
> Fedora builds. Apparently it does not find packages for the buildroot
> for i386.
>
> Is this a problem?
>
> http://koji.katello.org/koji/tasks?start=50&state=all&view=tree&method=all&order=-id
>
> LZ
>
> --
> Later,
> Lukas #lzap Zapletal
>
> _______________________________________________
> Pulp-list mailing list
> Pulp-list@redhat.com
> https://www.redhat.com/mailman/listinfo/pulp-list
I'd be OK with evaluating Copr, but wouldn't want to move until the
migration until a new Fedora Cloud instance has been made. Currently
the reliability's not great and the queues can get very long (hours, a
whole day). That's enough to make day to day work difficult, and would
be a real pain with releases.
My main concern apart from that is designing a release workflow with
Copr that makes it possible to put together releases from multiple Copr
repos and avoids rebuilding dependencies between releases. Nothing too
hard I shouldn't think, but clearly it won't be a 1-1 mapping between
Koji and Copr workflows.
We also need to keep the capability to build 1.6 releases in Koji over
the next few months, and probably 1.5 for a month or so.
···
On 03/09/14 10:07, Lukas Zapletal wrote:
> Decommissioning our koji is our priority in the upcoming weeks. We need
> to look at Copr and evaluate if it fits our needs. We are looking for
> public VM hosting with enough disk space (1 TB data volume) if Copr does
> not work for us - if you can help drop us a line.
