Well, I was wrong, /etc/init.d/glusterfs-server stop then start did solve
the problem. Guess it's really not the same problem as the one my colleague
on 3.8 is having :)
On Sun, Jan 15, 2017 at 03:06:54PM +0100, Kevin Lemonnier
wrote:> Hi,
>
> Our monitoring has been alerting about ram usage for a few days on
> our longest running proxmox / glusterFS cluster. Up until now
> I've just been migrating VMs to other hosts without paying much
> attention but I just realised that I have only 3 VMs left on the
> problematic server, and they really shouldn't be using that much ram.
>
> So I took a real look : glusterd is using 31,3 G of ram. That
> really wasn't the case at first, and isn't on the other members
> of that cluster (the two other replicas are using 2 G of ram).
>
> Two questions,
> 1) How do I safely free that ram ? I believe stopping gluster, at
> least on debian, doesn't actually stop the brick so the ram isn't
> freed. Is killing it safe ?
> Guess at worst I can migrate the remaining VMs over and just reboot
> the server, but I really really can't afford a problem right now.
>
> 2) What do you need to track down the leak ? It's 3.7.12, maybe
> it's already been solved ? I know a colleague is having a similar
problem
> on other servers with up to date 3.8, but they aren't doing VM hosting
> so maybe it's unrelated ..
>
> --
> Kevin Lemonnier
> PGP Fingerprint : 89A5 2283 04A0 E6E9 0111
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://lists.gluster.org/mailman/listinfo/gluster-users
--
Kevin Lemonnier
PGP Fingerprint : 89A5 2283 04A0 E6E9 0111
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: Digital signature
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20170115/65b09afc/attachment.sig>