I'm seeing this problem with both the virtio and e1000 driver with both linux and windows guests. About 5 - 10% of the VMs I launch have network problems. They seem to always get an IP from the DHCP server (dnsmasq), but afterwards either they can not make an outside network connection or they lose connectivity at a later time. Sometimes the connectivity comes back, sometimes it doesn't. It's very sporadic and there doesn't seem to be any rhyme or reason to it. The network is configured like "Routed network config" described here: http://libvirt.org/formatnetwork.html#examplesRoute The host machines are on Ubuntu 10.04. We were using the default Lucid packages (libvirt 0.7.5 / qemu 0.12.3), but I have upgraded us to libvirt 0.8.1 and qemu 0.12.4. I believe this has helped reduce the failure rate, but it still persists. Has anyone else had this problem or is there some suggestions for things to look into? Thank you. .:. jeremy -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://listman.redhat.com/archives/libvirt-users/attachments/20100809/17d30782/attachment.htm>
On Mon, Aug 9, 2010 at 1:13 PM, jeremy avnet <brainsik+libvirt at saucelabs.com<brainsik%2Blibvirt at saucelabs.com>> wrote:> I'm seeing this problem with both the virtio and e1000 driver with both > linux and windows guests. About 5 - 10% of the VMs I launch have network > problems.Interestingly, if I restrict the host machine to only ever allow one VM, the network problems go away completely. I just launched 75 VMs (one after another) and none of them had a problems. It's only if there are multiple VMs. Specifically, the network problems seem to occur when another VM is being launched. Maybe something is happening with the virtual bridge? .:. jeremy -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://listman.redhat.com/archives/libvirt-users/attachments/20100809/fb09d598/attachment.htm>