Re: [Ovirt-devel] iSCSI and disk timeouts

On Wed, Sep 15, 2010 at 4:12 PM, Justin Clacherty <justin redfish com au> wrote:

On 1/09/2010 8:00 AM, Justin Clacherty wrote:

On 1/09/2010 12:37 AM, Nicolas Ochem wrote:

Hi Justin,
I can suggest a method to make sure the problem is iscsi related :

-log in to the node
-copy the image file to RAM (by putting it in /tmp) or to an external storage if it's too big
-restart the VM directly in libvirt (virsh dumpxml, virsh destroy , virsh create) after having modified the XML file to point to the local version iso iSCSI

If the problem is gone then iscsi is the issue.
Hope this helps

Thanks Nicolas, I'll give that a go.

I've tried doing this and am still experiencing the problem so it seems iscsi isn't it. It does seem to coincide with disk access though. There are no messages on the guest when it dies, it just hangs. It looks as though the guest is crashing.

The following messages appear in dmesg on the node and seem to coincide with the guest hanging.

kvm: 19028: cpu0 unimplemented perfctr wrmsr: 0x186 data 0x130079
kvm: 19028: cpu0 unimplemented perfctr wrmsr: 0xc1 data 0xffd7695d
kvm: 19028: cpu0 unimplemented perfctr wrmsr: 0x186 data 0x530079
kvm: 19028: cpu1 unimplemented perfctr wrmsr: 0x186 data 0x130079
kvm: 19028: cpu1 unimplemented perfctr wrmsr: 0xc1 data 0xffd7695d
kvm: 19028: cpu1 unimplemented perfctr wrmsr: 0x186 data 0x530079

Is there a way to tell if it's the node causing the guest to hang?

Cheers,
Justin.