Newsgroups: comp.parallel.pvm
From: massimo@rios1.seas.ucla.edu ()
Subject: IBM SP2
Organization: IBM T.J. Watson Research Center
Date: 19 Jan 1996 15:07:09 GMT
Message-ID: <4doc2t$1lg8@watnews1.watson.ibm.com>

Are you using LL to submit the PVMe job? By default PVMe reserves the nodes
in dedicated mode, so even serial jobs might cause a node to be considered
"unavailable". You could try to add the option "-share cpu" when you start 
the daemon. This is described in the PVMe doc. 
Hope this helps,
Massimo
(massimo@vnet.ibm.com)

