Hello,

We are having trouble with a Solaris client having slow read performance. Both the Filer, F630 running 5.3.6R2, and Solaris, Ultra 2 running 2.6, client are on 100BT networks coupled thru Linksys EF24G2 Fast Etherswitch.

I looked at the obvious, all ethernet devices are in 100BT mode, running full duplex.

SUNW,hme0: Using Internal Transceiver SUNW,hme0: 100 Mbps full-duplex Link Up

slot 4: 10BaseT/100BaseTX Quad-Ethernet Controller e4a MAC Address: 00:c0:95:e0:2f:48 (100tx-fd-up)

Both are running V3 NFS over UDP. I set the filer to nfs.udp.xfersize=8192. (We changed this from the default value with only small improvement in performance.)

The filer is serving just a few clients and at this point the only real, demanding, client is this one Sun. Reads are quote slow.

Here's what the filer says during this time:

mlswfsa> sysstat CPU NFS CIFS HTTP Net kB/s Disk kB/s Tape kB/s Cache in out read write read write age 1% 0 0 0 0 2 42 42 0 0 >60 1% 0 0 0 0 2 45 52 0 0 >60 1% 0 0 0 0 1 27 27 0 0 >60 1% 2 0 0 1 2 15 14 0 0 >60 1% 0 0 0 0 2 43 52 0 0 >60 3% 18 0 0 3 52 197 83 0 0 >60

I do see a some timeouts and retrans on the Solaris box:

Client rpc: Connection oriented:

calls badcalls badxids timeouts newcreds badverfs 929582 9 1 8 0 0 timers cantconn nomem interrupts 0 0 0 1

Connectionless: calls badcalls retrans badxids timeouts newcreds 28489619 630 323405 5 324069 0 badverfs timers nomem cantsend 0 238992 0 0

Any pointers?

Thanks in advance.

Dan O'Brien, dmobrien@lcsi.net Cell: 614-783-4859 Work: 614-476-8473 Home: 740-927-2178 Pataskala, OH