I have had a sles10sp4 out-the-box install running for a long time - worked perfectly as a XEN host with several DomU’s.
I now upgraded host to sles11sp2 incl. all needed patches from online update. My DomU’s were backed up with dd command (from separate partitions to external disk) and restored back with dd to new linux raid1 partitions. Most seems OK, but I have a Windows XP DomU giving me headache: Shortly after bringing the DomU up (xm start) the host will crash completely - some kind of unclean forced shutdown, I am not sure what happens.
Restarting host it will recover file system from journals, similar to after a power failure. The host seems rock stable without this specific DomU.
anything in the logs (Dom0 syslog, Dom0 BIOS log, Dom0 xend.log)? If not - could you be a bit more specific on what you see? “Unclean shutdown” sounds like some shutdown, albeit incomplete, took place… or did you mean “crash”? How about opening an SSH session to Dom0 from a remote system and run “tail -F /var/log/messages” and maybe periodic “dmesg” in a second ssh session, so you’d potentially have some records surviving the crash?
We’ve been running WinXP on SLES11SP1 for quite some time, coming from SLES10SP4, too (and by now on 11SP2), without such problems. So it can be done
anything in the logs (Dom0 syslog, Dom0 BIOS log, Dom0 xend.log)? If not - could you be a bit more specific on what you see? “Unclean shutdown” sounds like some shutdown, albeit incomplete, took place… or did you mean “crash”? How about opening an SSH session to Dom0 from a remote system and run “tail -F /var/log/messages” and maybe periodic “dmesg” in a second ssh session, so you’d potentially have some records surviving the crash?
[/QUOTE]At least I haven’t been able to find the right hint, it might be there. Yes, it is a crash: Host dead with disk turned off, OS has experienced an unclean shutdown, only power off and back on is possible.
Any suggestions to what I might look for?
[QUOTE=jmozdzen;13059]
We’ve been running WinXP on SLES11SP1 for quite some time, coming from SLES10SP4, too (and by now on 11SP2), without such problems. So it can be done
[/QUOTE]Well, nice to know
[QUOTE=jmozdzen;13059]
Is the WinXP VM your only HVM on that Dom0?
[/QUOTE]Yes
If nothing is stored in the logs after the crash, then run a “tail” on these logs before the crash, from a remote system, so you have a chance to see if any messages were reported but not stored on disk (since the host crashes too fast).
At least I haven’t been able to find the right hint, it might be there.
Have you checked the logs I referenced, or was your reply of a more generic nature? Is there anything at all close to the time of crash in these logs?
Such sudden crashes usually leave a trace in the BIOS log (if this is server hardware). Especially since the kernel, to my experience, will either halt or reboot, but not power off the system. But dmesg, xend.log and the DomU’s log are the most promising places to find some hint.
