On 16.04.2013 10:34, Alex Leonhardt wrote: > Hi All, > > Am new to this list, and have asked in #ovirt / ovirt-users list and > they pointed me to here ... I've seen libvirtd using ~11 GB of resident > memory after the host has been running for about 60+ days with about 40 > VMs running on it. > > The versions I'm currently running are : > > libvirt.x86_64 0.9.10-21.el6_3.7 > libvirt-client.x86_64 0.9.10-21.el6_3.7 > libvirt-lock-sanlock.x86_64 > libvirt-python.x86_64 0.9.10-21.el6_3.7 > > I know I can just restart libvirtd and "all should be fine" (and > hopefully I'll regain the 11GB of memory) - however, is there an updated > version that maybe fixes a memory leak ?? Yes. John Ferlan did a great job in hunting memory leaks down. Most of his fixes went to 1.0.3, some into 1.0.4 or even 1.0.2. Michal

On 16.04.2013 11:29, Alex Leonhardt wrote: > Ah great, thanks! > > Alex In general, it can be a bit difficult to determine the exact commit which fixes problem you are seeing, because it depends on you concrete use case. However, you can try running libvirtd with valgrind and see where libvirtd leaks the most. This as disadvantage of libvirtd running a bit slower but on the other hand, if it is such huge leak even a little while should do. Maybe you will discover a new leak :) I got this trouble too,thanks.

2013/4/16 Michal Privoznik <mprivozn(a)redhat.com <mailto:mprivozn@redhat.com>> On 16.04.2013 11:29, Alex Leonhardt wrote: > Ah great, thanks! > > Alex In general, it can be a bit difficult to determine the exact commit which fixes problem you are seeing, because it depends on you concrete use case. However, you can try running libvirtd with valgrind and see where libvirtd leaks the most. This as disadvantage of libvirtd running a bit slower but on the other hand, if it is such huge leak even a little while should do. Maybe you will discover a new leak :) I met this issue.thanks