>It is common for us to experience very sluggish system performance and
>hung processes when we get the "NFS server (sever_name) not responding,
>still trying" message due to a mounted NFS system going down. (We are
>primarily using mounted NFS filesytems to move data files, usually with
>shell scripts.)
>I'd like to hear possible ways to reduce the impact of this error. I'm
>considering adding the bg option to my NFS mounts. How about increasing
>the timeo value? How about implementing CacheFS or Automounting?
>Suggestions?
I believe the bg option just helps when mounting.
You can use the "soft" option. That allows the NFS I/O operation to fail
and whatever was reading/writing can try to carry on somehow. But the
default is "hard" which makes I/O retry forever. On some systems, this can
cause complete lockup if the I/O blocks.
Worst of all is if you have hard cross-mounts (system A shares something
to B and B shares something else to A). In that case, meltdown on one
system can cause a domino effect that brings down the whole net, widens
the ozone hole, and causes 3 more species of lichen to go extinct.
Well, only kidding about the last two.
{cc to poster}
--
Secure webmail-news-voicemail-fax with JTAN ProShell
http://www.jtan.com/proshell - Unix SSHells for Pros
by