On Wed, Sep 24, 2014 at 02:58:59PM +0000, Daniel Helgenberger wrote:
On 24.09.2014 16:52, Sandro Bonazzola wrote:
Il 24/09/2014 16:14, Daniel Helgenberger ha scritto:
Hello Sandro,
this is the upstream 'working' bug, the original bug was flagged duplicate and posted for the 3.4 branch: https://bugzilla.redhat.com/show_bug.cgi?id=1124451 The bugs was opened against 3.4 and targeted to 3.6. Then it has been identified as a duplicate of a bug on ioprocess for 3.5 and fixed. Looks like assignee did not mark this for being backported to 3.4.z. Thanks for the info Sandro. IMHO this is a major issue and *should* be backported if possible to 3.4.x (right now I have a cron job restarting vdsmd once a week). I'll add a comment on BZ to make sure ;)
Toni can correct me, but I believe that he reprorted Bug 1124451 - vdsm leaks memory on a setup using ioprocess; that's why I've marked it as a dup of Bug 1130045 - Very high memory consumption Apparently, we have another memory leak, besides the one(s) instroduced in the 3.5 branch. It's not a new one - certainly not new to 3.4.4. I think it would be best to report a fresh one, with a precise vdsm version, and further info I requested in a parallel thread. Regards, Dan.