On Wed, Jul 02, 2014 at 03:15:40PM +0200, Michael Scherer wrote:
Le mercredi 02 juillet 2014 à 04:40 -0400, Anacron a écrit :
/etc/cron.daily/0logwatch:
gzip: stdout: No space left on device system zcat failed: 256 at /usr/sbin/logwatch line 896.
So, some cleaning was done this morning by Eyal and Ewoud on the server, and by me yesterday to get a few meg.
We are now at 2.4G, but discussing with Eyal on irc, we proposed to move the mailling list ( ie archive and stuff ) to a VM on alterway ovirt installation.
This may be disruptive, and since that's not in puppet, it may take longer than I wish. So I propose to install the services, but not declare them as being MX, do some testing.
Then once the lists are on the new list server, stop mailman on the current server, put it as secondary MX put the first one as primary MX and migrate over.
This should free 4G, but more importantly, this would split the download server from the ml, which mean that download server being full will not stop communication around the project anymore.
Any comment on this ?
Since I have no experience running mailman so I can't estimate all the risks, but you you should pay attention to IPv6. linode01 does have an AAAA record and at alterway we do not have IPv6. Should be fine because of CNAMEs, but can't hurt to check. Other than that I think we should upgrade alterway02. I think the oVirt version there isn't the most recent and any disruptive upgrades should be done before anything end-user facing runs there.