[Linux-cluster] Errors on lock server

[micah nerren]

On Mon, 2004-10-18 at 11:40, Michael Conrad Tadpol Tilstra wrote:
> On Mon, Oct 18, 2004 at 10:59:18AM -0700, micah nerren wrote:
> > Does anybody know what these messages mean (from /var/log/messages on a
> > lock_gulm server):
> > 
> > Oct 15 11:49:20 enigma kernel: lock_gulm: ERROR on action reply act:13 
> > err:1006
> > Oct 15 11:49:20 enigma kernel: lock_gulm: ERROR On lock 
> > 0x4703627374000905000000001e0de38800 fsid=bst: Got error 1006:Not
> > Allowed 
> > on Sync LVB request.
> [snip]
> > 
> > 
> > The file system is still accessible, but these are happening constantly.
> > Any help would be greatly appreciated!!
> 
> What version of the code?
> Have there been node failures?

The version is: GFS-6.0.0-7.1

There has been a node failure, or more precisely a failure of a file
system mounted on another node. Things were rebooted and all seems ok
except for those errors.


> Someone is trying to use LVBs and doesn't have the lock[s] in the correct
> state to do so.  This shouldn't ever happen. (we have seen this before,
> but I'm not remembering the conditions that caused it.)
> 
> With LVBs not working, gfs cannot cache any info there. (or the info is
> wrong)  This will affect df, quota, and posix range locks (on 5.2.* and
> 6.0.*) df and quota will be off, posix locks will probably panic.
> 
> 
> You really should shutdown/restart everything as soon as you can.  This
> is the easiest/quickest way to fix it.
> 

Ok we will try that. Is this a bug? Is there any info you need from the
system?

Thanks,

Micah