Re: All processes go tstile

["J. Hannken-Illjes" <hannken%eis.cs.tu-bs.de@localhost>]

> On 17. Aug 2018, at 04:46, Emmanuel Dreyfus <manu%netbsd.org@localhost> wrote:
> 
> On Thu, Aug 16, 2018 at 10:03:11AM +0200, J. Hannken-Illjes wrote:
>> Looks like a deadlock where we sync a file system, need a new buffer
>> and try to free a buffer on a currently suspending file system.
>> 
>> VFS_SYNC and VOP_BWRITE should be on different mounts here.
> 
> Here are the mounts:
> /dev/raid3a on / type ffs (log, local)
> /dev/raid3e on /home type ffs (nodev, noexec, nosuid, NFS exported, local)
> 
> The problem arises while dump is performing a snapshot backup on /home

The first thirty lines of "dumpfs /home" please.

Did you use "dump -x ..." or "dump -X"?

Did the "dump" process hang?

> It is worth noting that /home does not have -o log because frequent
> fsync from NFS clients kill performances with -o log. At least it did 
> it on netbsd-7, but I do not see why it could have changed on netbsd-8: 
> frequent fsync means frequence log flushes on the server.
> 
>> Do you have a crash dump?
> 
> I was not able to get one so far, the dump device is not properly
> configured on this system.

If it hits you again, from ddb:

ps /l
call fstrans_dump

backtrace of the "dump" process

--
J. Hannken-Illjes - hannken%eis.cs.tu-bs.de@localhost - TU Braunschweig (Germany)