Thanks to Tom Smith, and Johan Brusche for their responses, to do with
clua_services settings and inappropriate default routes following a
possible network failure, respectively.
In fact, the solution lay elsewhere, and was discovered by one of my
fellow admins here - some admin or process unknown had modified
/etc/ifaccess.conf to deny connections from the cluster machines to the
alt0 interface. Consequently, it couldn't hear its brother members.
ifconfig alt0 -filter
on all nodes, followed by reconfiguration of the ifaccess.conf and
reenabling the filtering fixed the problem.
Thanks again for all responses, as fast as ever.
Tim
--
Dr Tim Cutts
Informatics Systems Group, Wellcome Trust Sanger Institute
GPG: 1024D/E3134233 FE3D 6C73 BBD6 726A A3F5 860B 3CDD 3F56 E313 4233
Received on Tue Feb 22 2005 - 20:53:45 NZDT