[libvirt] disk XML for error policy

Hi Dan, What do you think of this for the disk XML specifying what to do on error: <disk type='file' device='disk'> <source file='/storage/guest_disks/libgdu'/> <target dev='vda' bus='virtio'/> <on_err rerror='stop' werror='stop'/> </disk> Dave

On Tue, Mar 16, 2010 at 10:14:26PM -0400, Dave Allan wrote:
Hi Dan,
What do you think of this for the disk XML specifying what to do on error:
<disk type='file' device='disk'> <source file='/storage/guest_disks/libgdu'/> <target dev='vda' bus='virtio'/> <on_err rerror='stop' werror='stop'/> </disk>
I'm in two minds as to whether we really need separate policies for read and write. I know QEMU supports that, but even those who implememted it told me it was rather redundant & a historical accident because they did not want to rename the existing 'werror' attribute when they added the 'rerror' bit. So unless someone has a compelling reason for them to be separate I think we can have a single policy. If we follow the style of the <on_reboot> <on_poweroff> elements we'd have something like <on_error>stop</on_error> My only other thought would be to do it as an attribute on the existing <driver> element in disks, where we have disk format & cache mode already Regards, Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

On 03/17/2010 11:09 AM, Daniel P. Berrange wrote:
On Tue, Mar 16, 2010 at 10:14:26PM -0400, Dave Allan wrote:
Hi Dan,
What do you think of this for the disk XML specifying what to do on error:
<disk type='file' device='disk'> <source file='/storage/guest_disks/libgdu'/> <target dev='vda' bus='virtio'/> <on_err rerror='stop' werror='stop'/> </disk>
I'm in two minds as to whether we really need separate policies for read and write. I know QEMU supports that, but even those who implememted it told me it was rather redundant& a historical accident because they did not want to rename the existing 'werror' attribute when they added the 'rerror' bit. So unless someone has a compelling reason for them to be separate I think we can have a single policy.
Heh, makes sense to me; one policy it is.
If we follow the style of the<on_reboot> <on_poweroff> elements we'd have something like
<on_error>stop</on_error>
My only other thought would be to do it as an attribute on the existing <driver> element in disks, where we have disk format& cache mode already
Adding an attribute seems like a nice way to do it. Dave
participants (2)
-
Daniel P. Berrange
-
Dave Allan