fsanta
Anyway, if you use hard mounts and clients give up then you have a bug in the NFS code. Which kernel do you use?
Ah. Now we're getting somewhere. I use 2.4.21, the one that comes with SuSE 9.0
Make sure you use the latest, for instance k_deflt-2.4.21-166
This only happens when we all try to login at the same time. If we stagger logins it's fine. This is what leads me to believe it's a nfs problem rather than dcop or kde.
Yes, it looks like an NFS problem. It may be due to an overloaded server or a kernel bug. Your description isn't precise enough to decide.
exports on the server: /home *(rw,no_root_squash)
Let the server operate in the asynchronous mode: /home *(rw,no_root_squash,async) It improves performance, see "man exports". Forget about the risk of data inconsistency, it makes sense in a production environment but not in a class which doesn't work anyway. Also, try to enable the write-caching feature of your IDE disk by "hdparm -W 1 /dev/hdX". It may also improve performance.
Here is an example the fstab entry on a client: 192.168.1.3:/home /home nfs rw,sync 0 0
IMHO the sync option is ignored here. Don't specify "soft" here, "hard" is the default that you want. "intr" may help when the server is not responding. You may also try the following options: noacl : to reduce NFS traffic tcp or udp : to test there isn't a problem with the implementation timeo : to reduce NFS traffic Start for instance with: 192.168.1.3:/home /home nfs rw,rsize=8192,wsize=8192,timeo=14,intr,noacl 0 0 -- A.M.