On Wed, Feb 26, 2014 at 04:52:14PM +0530, Pratik Mehta wrote: > On Wed, Feb 19, 2014 at 11:43 PM, David Teigland <teigland@xxxxxxxxxx> > wrote: > > > > > > That's a fine solution. You might also be able to use > > 'service cman start quorum'. > > Apart from DLM, wouldn't this prevent fenced from starting? Trying this > caused cman status to be non-zero (fenced is stopped), causing pacemaker > start to force a "cman start". If you don't need dlm or gfs, then you shouldn't need anything after the quorum breakpoint. It may be possible to insert your own step after the cman init script that stops or removes daemons and/or modules you don't need. > > The cman init script could probably > > use some sysconfig option to either disable dlm/gfs2/etc or to > > tell it to quit after the quorum step. > > I can help do this. Is there an equivalent example I can look at? Would you > envision this to be exported as a exclude-functionality type of config or > more as a breakpoint=foo (linear/procedural exclusion)? I'd probably use a config option for each start step. I'm not entirely sure if it would be taken into the cluster source tree at this point. > > Not long ago it was possible to avoid loading sctp, but people kept adding > > sctp symbols and I didn't have to time to try to keep them out. It would > > be nice if that could be corrected again. > > > > Any pointers on which release/timeline can I go look back at source from? I > can take a shot at replicating that. Hm, if you're using RHEL6 it would require working on the RHEL6 kernel, which is going to be different from upstream in the area of SCTP. And again, it's unlikely such a patch would be taken for RHEL6 at this point. Dave -- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster