1) The CentOS Linux server with all latest updates 2.6.32-358.14.1.el6.x86_64 (use fresh installation)<br />
2) There are 32 Windows XP 32 bit VM running on server with the same configuration: us01, us02,...,us32<br />
3) Every virtual machine has one it's own virtual network, <forward mode='route'/>: net01,net02,...,net32<br />
4) There is a script that control jobs running in every VM and do "virsh reboot VM" if needed, for example, if IE crashed<br />
<br />
Before latest updates the virtual machines works without problems: 24x7, all VM always running. Some days ago I found that multiple VM are in "shut off" state, but my script never do "virsh destroy VM" or virsh shutdown VM"<br />
<br />
Also I found the strange behaviour of libvirtd restart:<br />
<br />
sh-4.1# date ; virsh list | wc<br />
Mon Aug 5 16:34:03 MSK 2013<br />
33 94 1478<br />
sh-4.1# date ; /etc/init.d/libvirtd restart ; virsh list | wc<br />
Mon Aug 5 16:34:23 MSK 2013<br />
Stopping libvirtd daemon: [ OK ]<br />
Starting libvirtd daemon: [ OK ]<br />
32 91 1432<br />
<br />
As we can see one of the was killed<br />
The debug log of /var/log/libvirt is in the attachment
↧
