On Tue, 2010-01-19 at 19:09 +0000, Phillip Mumford wrote:
> Any idea on this? I've been having similar problems on a variety of machines:
This message indicates only a particular class of bug ('soft-lockup'),
not a specific bug. So unfortunately I cannot make any general
suggestions.
[...]
> I've had these problems with the 2.6.26 Lenny kernel and 2.6.30 from
> backports. These machines are fairly lightly used, but end up
> freezing occasionally, or in the worst case end up eating 100% CPU for
> several hours and require restarting. I've also tried 2.6.32 from
> sid, it also ended up with a task blocked within a day, so didn't
> bother let it running for longer to see if it crashed and burned -
> it's a production server.
Given that this is reproducible in 2.6.32, please submit a report
upstream at <http://bugzilla.kernel.org>.
> All the problems I've had are on virtual machines running on Xen hosts
> with more than 1 CPU core. CentOS 5.4 host server running Xen 3.0
> (plus plenty of additional RedHat patches probably), as well as a
> Linode server which is using Xen 3.4.1. Virtual machines running
> Lenny with just 1 core have no problems at all, however putting a
> Lenny virtual machine on a Xen host with access to more than 1 cpu
> core is almost guaranteed to end up with problems eventually.
[...]
The Xen kernel patch used in 'lenny' appears to be unstable in
multi-vCPU domains, and it seems unlikely that this will be solved now.
Note that the linux-image-2.6.26-686-bigmem (32-bit) kernel can also run
in domU and should be more stable.
Ben.
--
Ben Hutchings
Any smoothly functioning technology is indistinguishable from a rigged demo.
Attachment:
signature.asc
Description: This is a digitally signed message part