i''m crossposting from the opensolaris help forum: i''ve just migrated some zones to newer hardware running b94, and find that after a while working with an hvm in xvm, the machine hangs/panics. if i leave xvm stuff alone, the machine is perfectly reliable. it''s x64, intel chipset/procs anybody else having these troubles - zones and xvm together causing hangs/panics? cheers, blake This message posted from opensolaris.org
On Mon, Aug 04, 2008 at 08:39:31AM -0700, Blake Irvin wrote:> i''m crossposting from the opensolaris help forum: > > i''ve just migrated some zones to newer hardware running b94, and > find that after a while working with an hvm in xvm, the machine > hangs/panics. if i leave xvm stuff alone, the machine is perfectly > reliable. > > it''s x64, intel chipset/procs > > anybody else having these troubles - zones and xvm together causing hangs/panics?I''ve not seen this reported. Could you provide the stack traces?
How should I go about that? I''m just beginning to get into low-level debugging of this kind. I''m not sure what tools to use. Thanks, Blake This message posted from opensolaris.org
Blake Irvin wrote:> How should I go about that? I''m just beginning to get into low-level debugging of this kind. I''m not sure what tools to use. >When dom0 rebooted after a crash, there should be a core dump file saved in /var/crash/<hostname>/*.#. So, you can cd into that directory and type ''mdb #'' (replace # with the real number) to inspect the core. Once you get into mdb, type ''$c'' to get the stack trace. Max> Thanks, > Blake > > > This message posted from opensolaris.org > _______________________________________________ > xen-discuss mailing list > xen-discuss@opensolaris.org >
Oddly, there isn''t anything in the crashdump dir. What else can I try? cheers, Blake This message posted from opensolaris.org
Blake Irvin wrote:> Oddly, there isn''t anything in the crashdump dir. What else can I try? >You mean the machine panic''ed without saving any core dump file? Hmm...you may want to enable kmdb when you boot the machine (use ''mdb -K'' after you log in). Then, when it panics, it will drop into kmdb where you can type ''$c'' to see the back trace. Max> cheers, > Blake > > > This message posted from opensolaris.org > _______________________________________________ > xen-discuss mailing list > xen-discuss@opensolaris.org >
> Oddly, there isn''t anything in the crashdump dir. > What else can I try?Is the dump device (swap slice, or zfs dump zvol) big enough to store the kernel crash dump? This can be an issue after adding more physical memory to the machine, or when you did a manual layout of the disk slices and specified a swap slice size that is too small. Is there enough free space in /var/crash/`hostname` to save the crash dump? This message posted from opensolaris.org