I think Dan is right as I had the same kind of problem earlier. I fixed this issue by following a Dan's advice on another bug involving performance issue. You can find the fix in a previous message on this list:
http://www.redhat.com/archives/fedora-xen/2007-April/msg00166.html
The key is to mount /var/lib/xenstored in shared memory to limit som I/O overhead.
Maybe it should help you too.
Regards,
Guillaume
Le lundi 14 mai 2007 à 14:02 +0100, Daniel P. Berrange a écrit :
On Mon, May 14, 2007 at 02:40:01PM +0200, Frederic Saint-Marcel wrote: > Hello, > I have a Dell 2950 with : > > kernel-xen.i686 2.6.18-1.2798.fc6 > installed > xen.i386 3.0.3-8.fc6 > installed > xen-libs.i386 3.0.3-8.fc6 installed > > When I restart domU with dom0 reboot or /etc/init.d/xendomains script I > have this error > in my xend.log : > > [2007-05-14 14:20:07 xend.XendDomainInfo 7870] WARNING > (XendDomainInfo:882) Domain has crashed: name=xxxx id=21. > [2007-05-14 14:20:08 xend.XendDomainInfo 7870] ERROR > (XendDomainInfo:1675) VM xxxx restarting too fast (7.688173 seconds > since the last restart). Refusing to restart to avoid loops. > > on eight domU I have only six which restart normally and two with this > error. The xendomains script has not proved to be very reliable. I've seen a number of bugzillas with a similar complaint to you, that only a (random) fraction of guests successfully start upon boot. My best guess is that the high load causing by starting lots of guests at boot causing timeouts when setting up backend devices & thus crashes the guest. Not had any time to investigate it seriously, or even see if this guess is correct. Dan.
-- Fedora-xen mailing list Fedora-xen@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-xen
