I use the same Centos and Gluster versions and also see high load averages. I'm not sure I can help but I can relate since I have similar problems.
How many virtual machine images do you have on one brick? I started out putting several VM images on a single gluster volume and have since started creating a dedicated volume for each VM. My thinking was that each brick gets a single process on each peer and that having separate processes would prevent VMs from having to compete for slices of a single brick process. I have not experienced a noticeable difference between the setups so it is quite possible that my hypothesis was incorrect. What is the underlying hardware setup? I noticed significant VM performance improvements when I reconfigured the LVM LVs that backend my VM images from raid 1 to raid 10. High load average did not go away but VMs behave normally now instead of really slowly. How many CPU cores do you have? On my system, which is is a pair of ~5 year old supermicro servers with 8 cores and 8 GB of memory I consistently see load averages in the mid-20s. It's higher than I would like but the VMs that use the storage are still running well. -Mark On Mon, Dec 2, 2013 at 5:30 AM, gandalf istari <[email protected]>wrote: > I'm running version glusterfs 3.4.1 on a Centos 6.4 > > now Load average: 3.37 3.41 3.29 > > > the program that uses the most cpu: > /usr/sbin/glusterfsd but several instances. > > The logs are not showing any errors. > > > I have never used atop, but this looks like a problem > > LVM | s02-LogVol00 | busy 80% | read 0 | write 1166 > > gr > > > hi > which version of gluster > on which OS system version > Did you run "top" (or even better "atop") to see which programm uses up > how much resources (IO,RAM,CPU)? > anything unusual in the logfiles? > > > Am 02.12.2013 14:57:32, schrieb gandalf istari: > > Hi, I have two glusterfs server in replicated mode. > brick one has my virtual machines and brick two server receives only the > replicated parts. > > Now for the last 3 day's I get following message from my monitoring system. > > WARNING - load average: 6.04, 6.55, 6.07 > > And the load sometime goes on for about 16 hours. > > For me this seems not to be normal, anyone some advice on this ? > > The load on brick one is not more the 1.8 > > > Thanks > > _______________________________________________ > Gluster-users mailing list > [email protected] > http://supercolony.gluster.org/mailman/listinfo/gluster-users > > > > > -- > ------------------------------ > [image: *Ecologic Institute*] *Bernhard Glomm* > IT Administration > > Phone: +49 (30) 86880 134 Fax: +49 (30) 86880 100 Skype: > bernhard.glomm.ecologic [image: > Website:] <http://ecologic.eu> [image: | > Video:]<http://www.youtube.com/v/hZtiK04A9Yo> [image: > | Newsletter:] <http://ecologic.eu/newsletter/subscribe> [image: | > Facebook:] <http://www.facebook.com/Ecologic.Institute> [image: | > Linkedin:]<http://www.linkedin.com/company/ecologic-institute-berlin-germany> > [image: > | Twitter:] <http://twitter.com/EcologicBerlin> [image: | > YouTube:]<http://www.youtube.com/user/EcologicInstitute> [image: > | Google+:] <http://plus.google.com/113756356645020994482> Ecologic > Institut gemeinnützige GmbH | Pfalzburger Str. 43/44 | 10717 Berlin | > Germany > GF: R. Andreas Kraemer | AG: Charlottenburg HRB 57947 | USt/VAT-IdNr.: > DE811963464 > Ecologic™ is a Trade Mark (TM) of Ecologic Institut gemeinnützige GmbH > > _______________________________________________ > Gluster-users mailing list > [email protected] > http://supercolony.gluster.org/mailman/listinfo/gluster-users >
_______________________________________________ Gluster-users mailing list [email protected] http://supercolony.gluster.org/mailman/listinfo/gluster-users
