Description of problem: We've the problem, that when rebooting the Dom0 with "reboot", the DomUs are not shut down correctly, because of "XendError: /usr/lib64/xen/bin/xc_save 4 2 0 0 0 failed". And this results in, that the machines are not brought up after the reboot. Version-Release number of selected component (if applicable): xen-3.0.3-25.0.4.el5 How reproducible: Everytime, see above. Actual results: Unclean shutdown of DomU caused by a reboot at Dom0 results in no booting of DomU after the reboot (as in initial description). Expected results: Working stuff. Reboot of Dom0 cleanly shuts down the DomU and brings it also correctly back... Additional info: [2007-10-17 15:56:26 xend 4935] DEBUG (XendCheckpoint:83) [xc_save]: /usr/lib64/ xen/bin/xc_save 4 2 0 0 0 [2007-10-17 15:56:26 xend 4935] INFO (XendCheckpoint:239) Couldn't map p2m_frame_list_list (errno 14): 14 [2007-10-17 15:56:26 xend 4935] INFO (XendCheckpoint:239) Save exit rc=1 [2007-10-17 15:56:26 xend 4935] ERROR (XendCheckpoint:106) Save failed on domain central (2). Traceback (most recent call last): File "/usr/lib64/python2.4/site-packages/xen/xend/XendCheckpoint.py", line 100, in save forkHelper(cmd, fd, saveInputHandler, False) File "/usr/lib64/python2.4/site-packages/xen/xend/XendCheckpoint.py", line 227, in forkHelper raise XendError("%s failed" % string.join(cmd)) XendError: /usr/lib64/xen/bin/xc_save 4 2 0 0 0 failed [2007-10-17 15:56:26 xend 4935] DEBUG (image:458) hvm_shutdown fired, shutdown reason=halt [2007-10-17 15:56:26 xend.XendDomainInfo 4935] INFO (XendDomainInfo:899) Domain has shutdown: name=central id=2 reason=halt. [2007-10-17 15:56:26 xend.XendDomainInfo 4935] DEBUG (XendDomainInfo:1463) XendDomainInfo.destroy: domid=2 [2007-10-17 15:56:26 xend.XendDomainInfo 4935] DEBUG (XendDomainInfo:1471) XendDomainInfo.destroyDomain(2) [2007-10-17 15:56:26 xend 4935] DEBUG (image:447) hvm shutdown watch unregistered
Created attachment 229941 [details] Relevant parts from /var/log/xen/xend.log Please note, that this is a productive machine which I can't reboot just for fun or for just another test to see whether it works. Of course I'm willing to provide you any log details and further information you need, and I'll test an (inofficial) update also, if there's a chance, that it could solve this issue.
Correct me if I'm wrong; I believe the domains you are talking about are fully virtualized domains. If that is the case, the reason things aren't working is that fully virtualized save/restore doesn't work/isn't supported in RHEL 5.0. This should be addressed by 5.1, which does support fully virtualized save/restore. Chris Lalancette
Chris, do you mean there is no HVM support in RHEL 5.0 or is RHEL 5.0 just lacking the ability to save/restore HVM DomUs? Finally last of it would be IMHO just a logic feature by first. And first of it is supported as per RHEL 5.0 documentation. But to answer your question: Yes, the machine is fully virtualized. Anyway it would be strange to me to see SLES before RHEL regarding this. And wasn't RHEL 5.1 supposed to come up soon?
There is HVM support in 5.0; just not HVM save/restore. 5.1 is coming out fairly soon, although I don't have an exact schedule I can hand out. Chris Lalancette
Works with Red Hat Enterprise Linux 5.1 - closing now.
Can somebody please close this bug report? I'm unfortunately not allowed to do. It's fixed at least in xen-3.0.3-41.el5.