[users@httpd] how to isolate nfs subdirectories from killing Apache when nfs server is not responding.

[Tom H <to...@limepepper.co.uk>]

HI all,

I have an httpd-2.2.3 server running on centos5.4 x86_64

I have a couple of nfs mounted subdirectories like so;
xx.xx.xx.xx:imagesXXX on /var/www/html/images/imagesXXX type nfs4 
(ro,soft,timeo=10,retrans=0,proto=tcp,intr,addr=xx.xx.xx.xx)

Every so often the nfs server goes away, this has been due to hardware 
failure, disk corruption, or human error.

However despite trying lots of different mount settings, including 
hard/soft nfs/nfs4, tcp/udp, various timeo/retrans, if the nfs server is 
down, then the nfs client blocks on the request for at least a minute - 
(even soft, timeo=0, retrans=0, intr)

As there is reasonable load on the server, very quickly all the 
available client connections are blocked on the nfs request, and 
performance goes bad.

Are there any strategies to isolate the /var/www/html/images/imagesXXX 
so apache does not wait for more than say 5 seconds for the filesystem 
to return?

(i don't want to make any system wide changes, say to tcp, apache or 
system IO settings, as apache is running perfectly well in all other 
regards)

Many thanks
Tom


# cat /etc/redhat-release
CentOS release 5.4 (Final)
# uname -a
Linux serverXX 2.6.18-164.11.1.el5 #1 SMP Wed Jan 20 07:32:21 EST 2010 
x86_64 x86_64 x86_64 GNU/Linux
# rpm -qa | grep httpd
httpd-2.2.3-43.el5.centos

---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
   "   from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org