6-STABLE snapshot (background fsck) lock-up

[Eric Anderson]

On 08/17/06 10:04, Kostik Belousov wrote:
> On Thu, Aug 17, 2006 at 09:53:28AM -0500, Eric Anderson wrote:
>> On 08/17/06 07:30, Eric Anderson wrote:
>>> On 08/17/06 07:25, Kostik Belousov wrote:
>>>> On Thu, Aug 17, 2006 at 07:08:31AM -0500, Eric Anderson wrote:
>>>>> I've seen this several times now, but this time I got a dump. 
>>>>> Basically, the system comes up after unclean shutdown, throws a bunch of 
>>>>> filesystems into the background fsck list, and begins 60 seconds later. 
>>>>>   After some amount of time, the system will lock up, presumably hung 
>>>>> waiting on locks.
>>>>>
>>>>> Here's some info:
>>>>> lock type bufwait: EXCL (count 1) by thread 0xffffff0128812980 (pid 804) 
>>>>> lock type getblk: EXCL (count 1) by thread 0xffffff0128812980 (pid 
>>>>> 804)Locked vnodes
>>>>>
>>>>> 0xffffff01c5c9f000: tag ufs, type VREG
>>>>>    usecount 1, writecount 0, refcount 13656 mountedhere 0
>>>>>    flags (VV_SYSTEM)
>>>>>    v_object 0xffffff01b09542a0 ref 0 pages 1691637
>>>>>     lock type snaplk: EXCL (count 1) by thread 0xffffff0128812980 (pid 
>>>>> 804)
>>>>>        ino 4, on dev label/vol4
>>>>>
>>>>>
>>>>>
>>>>> How can I use kgdb to get more debugging information from this?
>>>>>
>>>> Look at the
>>>> http://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kerneldebug-deadlocks.html
>>>> for instructions for ddb.
>>>>
>>>> What is the date of the system checkout ?
>>>
>>> 6-STABLE from ~8/12/06.
>>>
>>> So there's not much that can be done with the vmcore?
>>>
>>> I'll add the options to my kernel config (I was missing a few), and I 
>>> suppose wait for another deadlock.
>> Ok, had another one.  Here are screenshots:
>>
>> http://www.googlebit.com/freebsd/snapshots/
>>
>> Is it the sysctl that is triggering it?
> 
> Hmm, unbelievable. It would be much simpler for you to set up serial
> console.

Not really, at least right now.  I'm doing this remotely, so it'll be 
some time before I can do that.

> Anyway, you use g_journal, and the problem seems to be related to some
> geom locking. pjd@ is the right person to contact. I added him to the Cc:.

Ahah.  Ok.  Thank you!

Eric


-- 
------------------------------------------------------------------------
Eric Anderson        Sr. Systems Administrator        Centaur Technology
Anything that works is better than anything that doesn't.
------------------------------------------------------------------------