hello managers,
I have a lot of NFS problem with my NFS server & clients.
I'll describe my conf first ..
The servers are Alpha 4100 5/300, with 128MB Memory & single CPU, in a
Trucluster. They run DU4.0B & TCR1.4. They have shared storage connected
using KZPSA controllers & RZ29B drives in BA356 shelves.
The clients are Alpha 500/333 workstations running DU4.0B with 64MB
memory. The network is a fully switched 10/100Mb with UB GeoLAN/500
switches. We are currently operating the servers as well as the clients
in full-duplex mode. Clients operate at 10 Mb, while servers operate at
100 Mb.
I have NFS services defined on the servers, which are the login & user
directories for all clients. Of these, 15-20 clients are active at a
given point in time.
Initially, the number of NFS daemons was 25 ( both TCP & UDP ). We
observed NFS write errors occuring infrequently from all clients.
However, the response at clients appeared to be slow. We increased the
number of daemons to improve this. This only increased the number of NFS
errors reported. Even a simple ls, cp or mv command on a NFS directory
would fail under these conditions, reporting an RPC timeout.
I reduced the number of daemons & observed a corresponding decrease in
the number of errors. Currently, only 10 daemons are running. However, I
still observe errors from all clients. 4-5 errors at least are observed
every hour, which increases with NFS activity of the client.
I have observed two errors. These are
Jul 29 12:04:23 hydra vmunix:NFS3 write error 60 on host nfs1cdata
Jul 29 12:04:23 hydra vmunix:NFS3 RFS_WRITE failed for server
nfs1cdata:RPC:timed out
Jul 29 14:33:32 hydra vmunix:NFS3 write error 70 on host nfsmission
nfsstat does not report any NFS server or client errors at all.
Is this normal or is there a real problem. If any one of u have faced
this problem and
have a solution pls help me out.
TIA
Umapathy J
Received on Thu Jul 30 1998 - 05:23:38 NZST