On Fri, 2011-04-22 at 21:18 -0300, Daniel Bareiro wrote: > On Friday, 22 April 2011 20:16:11 -0300, > Daniel Bareiro wrote: > > > > Were you doing anything when it hung or was the system just running > > > in a steady state? > > > Well, 35 minutes ago I had another hang. Same symptoms: connecting > > directly a keyboard and monitor, there is no reaction when pressing any > > key (black screen); loss of connectivity to the dom0 and domUs. > > > > Additionally I can comment that this time the crash occurred when I was > > doing an aptitude update/upgrade in the dom0 and the domUs. I can also > > comment that the disk activity LED was constantly on (not flashing). > > > > After hard reset it seems that something was unconscious in one of the > > RAIDs, and is currently syncing. > > > > In /var/log/messages there is no evidence of any particular event from > > the previous boot. The same in kern.log and syslog. If the disk controller has gone funny it pretty likely that whatever log messages were generated never made it onto the disk. > Here I found a thread [1] with a problem quite like this (although in my > case the hardware is different: A8V-MX motherboard and AMD Athlon 64 > Processor 3500+) in the xen-users mailing list. And another thread [2] > which is derived from the above. > > I'll try using "cpuidle=off" or "max_cstate=1" at xen cmdline in > /boot/grub/grub.conf. It's probably worth trying cpuidle=off but it looks like the max_cstate=1 thing is Intel specific. It's not clear to me in either case if the root cause of the issue this fixes is a h/w or s/w issue. Do you have irqbalanced installed/running? Ian. -- Ian Campbell Questionable day. Ask somebody something.
Attachment:
signature.asc
Description: This is a digitally signed message part