[Linux-cluster] GFS lock problem?
Andrew C. Dingman
adingman at cookgroup.com
Thu Jun 9 03:52:46 UTC 2005
Wouldn't each lock in GFS be used to impliment some sort of
filesystem-level lock, such as an fcntl() or flock(), which would in
turn belong to a process?
On Tue, 2005-06-07 at 13:39 -0500, Michael Conrad Tadpol Tilstra wrote:
> On Tue, Jun 07, 2005 at 01:12:51PM -0500, Andrew C. Dingman wrote:
> > Using GFS 6 on RHEL3, is there any way to determine which locks are held
> > by a particular node, or better yet by a process on that node?
>
> send a USR1 signal to the LT process, this will dump a snapshot of the
> entire lockspace into /tmp/Gulm_LT_Lock_Dump.LT000.?????? (the ? will be
> random chars.)
> `gulm_tool getstats localhost:lt000 | grep pid` will tell you the pid
> of the LT000 process.
>
> There isn't a way to see what processes hold locks because processes
> don't hold locks. GFS holds locks, not processes.
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> http://www.redhat.com/mailman/listinfo/linux-cluster
More information about the Linux-cluster
mailing list