slow nfs response with 5310
We have been seeing some very slow NFS response to our two 5310's. The problem increases as we get higher levels of activity. Doing a simple 'ls' in an NFS mounted directory will sometimes hang. No errors noted on 5310. No NFS errors reported by clients.
Anybody have any thoughts on the subject? Does the 5310 have the equivalent of nfsd daemons that can be increased? Does the 5310 have any network options that can be tweaked? With most of the files being on the smaller side (mb vs the gb range), would it help to force the rsize and wsize on the client NFS mount options? If so, anybody have a recommended size for these?
What we are using the 5310's for:
- automounted home directories (not more than 30-35 users at any given moment)
- data directories for Oracle (9 and 10)
- log file directory
- web server data directory (small web site, mostly static pages)
- shared directory for PC users
How we are set up:
- Two 5310's with 3 expansion units each
- Each EU split into two LUNs
- Numerous volumes created, some spanning LUNs
- Volume sizes range from 100GB to 1TB
- checkpoint turned off
- attic turned off
- all volumes exported to UNIX clients
- all volumes shared to PC clients
- authenticating to NIS+ server
- authenticating to WINS
- NTP enabled, sync'ing to local Unix box
- 4092 MB of memory in each 5310
- 5310's attached via single RJ45 (CAT-5e) to Cisco 4503 switch, running at 1000MB full duplex
- Unix clients are Sun V890s, attached via single RJ45 (CAT-5e), running at 100MB full duplex
- V890s running Solaris 9
- 5310's running v4.11 build 12
- No other network devices between 5310's and V890's.
thanks for any help on the problem...
Leif
