On 04.03.2022 10:21, BERTRAND Joël wrote:
Dima Veselov a écrit :
On Thu, Mar 03, 2022 at 12:17:32PM +0000, Stephen Borrill wrote:
On Wed, 2 Mar 2022, Dima Veselov wrote:
I have two similar NetBSD boxes, one is under 9_STABLE and the other is
old 8_STABLE. Sometimes one or another or even both get hit by something
and OS freezes. It looks like hardware stops working, but OS do not
catch panic. It means if ping is constantly running on console it would
look like network went down. Any command or logging in would hang
forever
like it happens when disk do not respond.
Providing a dmesg and information about things like firewalling would
be a good start. I've seen things like this when attempting to use
multiple bge NICs and also with IPFilter.
There is no ipfilter nor npf running. One system use software RAID and
the other have hardware RAID. Both have dynamic network configuration
using net/zebra, as we use it on many other systems. dmesg follows.
I see the same deadlock for a while on 8 and 9. LOCKDEBUG does provide
usable information.
Thanks for a hint. I have recompiled 9-STABLE kernel with DEBUG and
LOCKDEBUG. Is it enough to run server with serial console or I can get
something useful from GDB when the freeze will happen?
--
Dima Veselov
Physics R&D Establishment of Saint-Petersburg University
