According to the comments on launchpad, the bug concerns querying of
interfaces. I am using both bridged interfaces and virtual networks in
my setup. I also use isolated virtual networks, in which the host
machine does not participate, a feature virt-manager does not support,
yet still attempts to query (it goes "Unsupported data type: <type
'NoneType'>" when attempting to display it in the GUI).
Regardless, from the bugreport I linked to, it would seem others are
having this issue as well, and when it happens it's quite bad. A
megabyte a minute adds up if virt-manager is left running for a weekend.
On 13/05/14 13:51, anonym wrote:
[Note: I'm not subscribed, so please keep me CC:ed if you have any
follow-up questions.]
When virt-manager is connected to libvirtd from wheezy-backports, the
memory usage (both res and virt) of libvirtd on the server goes up by
about a megabyte a minute. It will keep going and eat several
gigabytes of RAM if left alone. Disconnecting virt-manager stops the
memory usage from increasing, but does not lower it back to an
acceptable level.
I'm using libvirt from wheezy-backports (1.2.1-1~bpo70+1), and I do not
experience this memory leak even after the same virt-manager (from
Wheezy, 0.9.4-2) process has been connected to the same libvirtd process
for at least two days.
The bug seems to be identical to the one described here, which is
resolved by using a newer version of netcf with libvirtd:
https://bugs.launchpad.net/ubuntu/+source/netcf/+bug/1201938
My netcf and libnetcf1 are from Wheezy (0.1.9-2).
Cheers!