[Linux-cluster] issues starting services

[Lon Hohberger]

On Fri, Jul 06, 2007 at 12:56:21PM -0400, Lon Hohberger wrote:
> On Mon, Jul 02, 2007 at 10:26:59AM -0500, Chris Harms wrote:
> > Had a nice little hardware failure over the weekend.  After having the 
> > machine come back on-line, 2 of the registered services didn't start 
> > (the fact that they weren't running already is a function of services 
> > not failing over until fencing succeeds).
> > 
> > Issuing a start operation in Conga did nothing. 
> > 
> > Issuing  clusvcadm -e [service] -m [node] yielded:
> >    Member [node] trying to enable service:[service]...Success
> >    service:[service] is now running on [node]
> > This was not the case.  Nothing happened.  Nothing was logged.

Oh - also, run:

rg_test test [your_cluster_conf] &> rg_test.out

If you have resource conflicts (e.g. the same IP referenced by two
services), the second service will have those resources ignored.

If you see something like "Primary/unique not unique...", you've got one
:)

-- Lon

-- 
Lon Hohberger - Software Engineer - Red Hat, Inc.