Hi
> I'm currently running a set of SLES10 SP2 servers that share common
storage using OCFS2. The VMs run fine, although sometimes it seems that the
disk files don't close out correctly and I end up with loop devices that
won't detach correctly. The bigger issue, though, is that if I try to do an
"xm save" on a domain and put the file onto the OCFS2 volume, the save fails
after a couple of seconds and a few bytes of data. If I save the file to
the local filesystem (ext3 or rieser) it works fine, but a save to the OCFS2
volume doesn't seem to work. Also, once the xm save fails, the domain gets
into a weird state that an "xm restore" or "xm resume" or "xm unpause"
doesn't seem to help. Then I have to reboot the physical server because the
qemu-dm process goes into a "defunct" state and won't close out and release
the file. Unfortunately I didn't get log files the last time this happened,
but I'll try to capture that the next time I see the behavior. In the
meantime, if anyone else has any ideas, I'd appreciate the input.
If you have a ?qemu-dm?-process, you are propably using HVM domains. I think
this is a bad idea, because:
- SLES10 has support for PV-domains in it?s kernel (AFAIK)
- PV is generally faster than HVM
- Suspend works reliable for PV-domains
- Depending on your very exact xen version, HVM don?t work reliable with
suspending.
So, try a PV-setup, if possible.
Regards,
Steffen
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
Home