[Linux-cluster] DLM messages

[Bas van der Vlies]

David Teigland wrote:
> On Mon, Mar 27, 2006 at 08:23:40AM +0200, Bas van der Vlies wrote:
>> GFS: cvs STABLE
>>
>> I just noticed that as we get these dlm messages the load will  
>> increase about 60 and stays there. what does this messages
>> mean?
>>
>> process_lockqueue_reply id 237030a state 0
> 
> These are common and haven't been associated with any problems
> before.  We should probably remove the printk.  It's caused
> by a grant message arriving before the reply to the lock request.
> 
Thanks i also found that answer in the archive

>> Mar 26 15:35:25 ifs2 kernel: dlm: lisa_vg3_lv1: cancel reply ret 0
>> Mar 26 15:35:25 ifs2 kernel: lock_dlm: unlock sb_status 0 2,15a940b8  
>> flags 0
> 
> I've never seen these before.  They're related to cancels which
> GFS only does during recovery.  What applications are using GFS?
> 
Our setup is a 4 node GFS-server that act as Home NFS-server for our 
cluster. This cluster is used by different universities and research 
institutes. All applications are using NFS and therfore also GFS.

Regards


-- 
--
********************************************************************
*                                                                  *
*  Bas van der Vlies                     e-mail: basv at sara.nl      *
*  SARA - Academic Computing Services    phone:  +31 20 592 8012   *
*  Kruislaan 415                         fax:    +31 20 6683167    *
*  1098 SJ Amsterdam                                               *
*                                                                  *
********************************************************************