Please keep all responses on the mailing list. I haven't heard about this problem before but I agree that it sounds like it's not related to OVS. It could be a bug in the Xen tools for Windows. If I recall correctly, some versions incorrectly set up the descriptors for TSO frames. You might be able to work around it by disabling large send offload in the guest.
On Wed, May 16, 2012 at 3:23 PM, Guilherme M. Schroeder <[email protected]> wrote: > I'm using version 1.4.0. I've compiled from master too, 1.7.90, same problem. > We still don't know what triggers it but it's related to Windows VMs. > We have Windows and Linux XenServer hosts and this problem just occur > on the Windows hosts. > The Windows hosts runs 2003 and 2008 versions. > > I've disabled "rx tx sg tso ufo gso" with ethtool on network > interfaces of dom0 and pif and vifs, but it still crashes, thought i > didn't get the dump from domain0 (it was empty). > > I'm afraid that this is a bug on kernel (2.6.32). We have other hosts > that run XenServer 5.6 in bridge mode and some days ago i got a stack > trace showing a crash on skb_gso_segments, so i don't think it's > related to OpenvSwitch. > > Right now i'm trying to setup a Debian Wheezy with xcp-xapi. It have a > newer kernel (3.2.0) also OpenvSwitch is buit-in. Gonna see how it > works. > > Let me know if you need more info or have some idea. > > Thanks. > > On Wed, May 16, 2012 at 2:27 PM, Jesse Gross <[email protected]> wrote: >> On Mon, May 14, 2012 at 12:50 PM, Guilherme M. Schroeder >> <[email protected]> wrote: >>> We have a lot of hosts that reboot randomly and today, using a new >>> hardware (HP BL460c G8) we got XenServer 6.0.2 to dump it's crash >>> successfully. >>> Kernel log is attached. >> >> What version of OVS is this? Is there anything in particular that >> triggers this? _______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss
