[Linux-cluster] Unable to obtain lock

Lon Hohberger lhh at redhat.com
Mon Jan 29 18:44:46 UTC 2007


On Fri, 2007-01-26 at 19:28 +0100, Marc Grimme wrote:
> On Friday 26 January 2007 19:15, Lon Hohberger wrote:
> > On Fri, 2007-01-26 at 09:19 +0100, Marc Grimme wrote:
> > > Hello,
> > > yesterday we saw a clusterfreeze (which seems to come from the rgmanager)
> > > with RHEL4/U4 GFS installed (see logs) consisting of 6 nodes x86_64
> > > Architecture. After fencing one node the cluster came back to live.
> > > Any idea what could have happend?
> >
> > Check 'dmesg' and 'cman_tool status'.  Also look at /proc/slabinfo,
> > specifically 'dlm_lkb' bits.  There's a chance that you hit a bug that's
> > already fixed. :)

> dlm_lkb           189628 195177    232   17    1 : tunables  120   60    8 : 
> slabdata  11481  11481    384
> nodea
> dlm_lkb           2074114 2077587    232   17    1 : tunables  120   60    8 : 
> slabdata 122211 122211    180
> nodeb
> dlm_lkb           454319 499392    232   17    1 : tunables  120   60    8 : 
> slabdata  29376  29376      0
> nodec
> dlm_lkb           242144 251719    232   17    1 : tunables  120   60    8 : 
> slabdata  14807  14807    480
> noded
> dlm_lkb           248672 286382    232   17    1 : tunables  120   60    8 : 
> slabdata  16846  16846    212
> nodef
> dlm_lkb            62934  62934    232   17    1 : tunables  120   60    8 : 
> slabdata   3702   3702      0

You've hit "the bug".

> > Need above information (and possibly more) to answer this.
> What more?? ;-)

Nothing; test packages here:

http://people.redhat.com/lhh/rgmanager-1.9.54-2.218112hf.i386.rpm
http://people.redhat.com/lhh/rgmanager-1.9.54-2.218112hf.x86_64.rpm
http://people.redhat.com/lhh/rgmanager-1.9.54-2.218112hf.src.rpm





More information about the Linux-cluster mailing list