On 08/05/15 13:12, Trevor Gale wrote:

> Awesome, I’ll look into that all. What do you mean by exports some scratch 
> space?

Scratch space is usually a large area of high performance disk that
isn't backed up and it designed for temporary use.

The problem is that NFS and high performance don't usually go together,
which is partly down to NFS (at least up to and including NFSv3) and
partly down to underlying filesystems.  For instance ext3 and ext4 used
to be (and probably still are) single-threaded through the journal
daemon and so if you have lots of concurrent writers then you can find
them all backing up whilst the journal daemon tries to keep up.  XFS can
be a good choice for that instead.

If you'll forgive a quick plug for another mailing list, there is a
mailing list for Linux clusters called the Beowulf list:

http://beowulf.org/

which has some good people with lots of experience in this and isn't
very high traffic these days.

Drop me a line if you'd like to be subscribed (had to disable Mailman
web subscriptions as it was being abused to mailbomb people).

All the best,
Chris
-- 
 Christopher Samuel        Senior Systems Administrator
 VLSCI - Victorian Life Sciences Computation Initiative
 Email: [email protected] Phone: +61 (0)3 903 55545
 http://www.vlsci.org.au/      http://twitter.com/vlsci

Reply via email to