I admin a cluster of ~30 2x & 4x PMacs (~90cores total). These
critters drive our large tiled display and also are used to do some
purely computational jobs as well.
Just yesterday (post time change, but pre-10.4.9 update) we were about
to do a big demo to some important guests and our AFP-mounted disks
(exported from a 10.4.8 server) started timing out. Since these disk
mounts were responsible for feeding apps to the cluster, the demo was
largely a failure (saved only by a post-doc who had an alpha-stage,
but non-AFP app ready to go).
After the guests left, we did a postmortem and after some debugging,
and after trying a number of things, in frustration we did the 10.4.9
update on the server - that by itself didn't help. Then we did the
10.4.9 update on the clients and that seems to have solved it.
If this is a wide-spread failure, you've already noticed it. If you
haven't upgraded to 10.4.9, you might consider doing it, on top of
the security considerations.
The other thing that we've noticed (re: using Macs in a cluster) is
that Mac's NFS from a Linux or Mac filer is (IMHO) just awful. It's
unusably unstable. Using SMB is more stable, but still hardly high
performance. The AFP is the most stable (/extremely/ stable to/from
a OSX server), but still not particularly zippy.
Just my 2c.
--
Harry Mangalam - Research Computing, NACS, E2148, Engineering Gateway,
UC Irvine 92697 949 824 0084(o), 949 285 4487(c)
email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Scitech mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/scitech/email@hidden
This email sent to email@hidden
