S. Ercan Yüzbasioglu
2009-Feb-20 11:29 UTC
[Xen-users] Paravirtualized DomU clock stops after migrating back
Hi all, I think this problem is well known, but I cannot say the same for the solution.. A similar issue was discussed in this mailing list: http://lists.xensource.com/archives/html/xen-users/2008-08/msg01039.html There are also two bug reports about it, which are marked as resolved, somehow... https://bugzilla.redhat.com/show_bug.cgi?id=426861 http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=1282 However, I cannot solve the problem using the workarounds that are discussed on all those threads. My problem differs slightly from the ones above: 1. I start the DomU on node1. Its clock works. I migrate the domu from node1 to node2, I open a domu console on node2 and everything is fine. Issuing "date" always gives the right time. However, when I migrate it back to node 1, and open console to domu, I observe that the clock stopped working. "date" gives always the same time, "sleep" won''t return, etc.. When I migrate it back to node 2, clock starts working again! 2. This happens after a cold migration too, not necessarily a live one. I think I did everything recommended on several forums. e.g.: - On the domu, I set the indepentent wallclock to 1 - I installed ntp client on DomU, I added a couple of ntp servers to ntp.conf (i.e. 0.pool.ntp.org) and ran ntpd before migration. - Both nodes are syncronized with the same ntp server. Nothing helped so far.. am I doing something wrong? Please help if you have any clue on this. Cheers Ercan _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
S. Ercan Yüzbasioglu
2009-Feb-20 11:58 UTC
[Xen-users] Re: Paravirtualized DomU clock stops after migrating back
Ok, looks like It''s been solved. Adding "clock=jiffies" as a kernel parameter in the DomU configuration did the trick. Time still slews more than 30 seconds after every migration though, and issuing "ntpd -q" with ~10 second intervals corrects the time ~0.01 sec every time. Not really accurate, but the most important thing is that it works! Actually I collect some measurements about resource consumption from those DomUs, so having an accurate clock is pretty important for those calculations. The DomUs are not supposed to have an access to the Internet, so I will have to install my own NTP server. If someone knows a better workaround for this problem to save me from these boring tasks, I will appreciate it! Cheers Ercan On Fri, Feb 20, 2009 at 12:29 PM, S. Ercan Yüzbasioglu < ercan.yuzbasioglu@gmail.com> wrote:> Hi all, > > I think this problem is well known, but I cannot say the same for the > solution.. > > A similar issue was discussed in this mailing list: > http://lists.xensource.com/archives/html/xen-users/2008-08/msg01039.html > > There are also two bug reports about it, which are marked as resolved, > somehow... > https://bugzilla.redhat.com/show_bug.cgi?id=426861 > http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=1282 > > > However, I cannot solve the problem using the workarounds that are > discussed on all those threads. > > My problem differs slightly from the ones above: > 1. I start the DomU on node1. Its clock works. I migrate the domu from > node1 to node2, I open a domu console on node2 and everything is fine. > Issuing "date" always gives the right time. However, when I migrate it back > to node 1, and open console to domu, I observe that the clock stopped > working. "date" gives always the same time, "sleep" won''t return, etc.. When > I migrate it back to node 2, clock starts working again! > > 2. This happens after a cold migration too, not necessarily a live one. > > > I think I did everything recommended on several forums. e.g.: > > - On the domu, I set the indepentent wallclock to 1 > - I installed ntp client on DomU, I added a couple of ntp servers to > ntp.conf (i.e. 0.pool.ntp.org) and ran ntpd before migration. > - Both nodes are syncronized with the same ntp server. > > Nothing helped so far.. am I doing something wrong? Please help if you have > any clue on this. > > Cheers > Ercan > > > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Reasonably Related Threads
- Error on xm create: VmError: (38, ''Function not implemented'')
- strange hostname issue on volume create command with famous Peer in Cluster state error message
- megatec process die
- strange hostname issue on volume create command with famous Peer in Cluster state error message
- strange hostname issue on volume create command with famous Peer in Cluster state error message