[Users] (quick tip) Controlling glusterfsd CPU outbreaks with cgroups

[Andrew Lau]

Hi all,

I've seen quite a few posts with people using similar scenarios of running
gluster and ovirt on the same physical box so I think this may be relevant.
After some discussions on the gluster mailing list, I put together a quick
post [1] about how you can limit glusters CPU resources so when it comes to
self-healing etc. your VMs will have higher priority and not crawl to a
halt when glusterfsd takes all the resources.

So far, I haven't been able to apply these cgroup rules without losing the
dynamic ones set by libvirt so you must do this in maintenance mode
followed by a full host restart after following those instructions.

I'm open to suggestions and feedback.

Thanks,
Andre
w

[1]
http://www.andrewklau.com/controlling-glusterfsd-cpu-outbreaks-with-cgroups/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20140204/fc9267a8/attachment-0001.html>