[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Qemu-discuss] QEMU process memory footprint

From: Vladislav Bogdanov
Subject: [Qemu-discuss] QEMU process memory footprint
Date: Wed, 21 Aug 2013 14:46:20 +0300
User-agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130625 Thunderbird/17.0.7

Hi list, newcomer is here.

I noticed unexpectedly big usage ( >120Mb, up to 200) of Rss by some
qemu processes, having all of them configured to use hugepages for RAM

Others with completely similar command line consume less than 80Mb.

Interesting enough, if I live-migrate VMs to another host, their Rss
footprint becomes much lower (~25Mb).

I run qemu-kvm 1.2.2 on Centos6, managed by libvirt-1.0.2. Intel CPUs,
KVM (from EL kernel).

I do not know, is it related or no, but VMs are usually started in
parallel by pacemaker CRM (Cluster Resource Manager), and if I restart
VM with big Rss footprint after some time on the same cluster node, Rss
size is much less too (~38Mb).

To summarize, in some cases I may have Rss
196Mb after "mass-start"
40Mb  after restart
25Mb  after migration

I believe, this could be considered as a bug, f.e. some rather big pages
used for initialization are not freed after they are not needed, and
there could be some race between processes being started. Although I do
not know, does it belong to qemu or to libvirt.

Is that a known issue that was already addressed?

Thank you,

reply via email to

[Prev in Thread] Current Thread [Next in Thread]