NFS client hanging

• Previous message: Bill Bradford: "ADMIN NOTE: Search engine back up"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

To: "'sunmanagers@sunmanagers.org'" <sunmanagers@sunmanagers.org>
Date: Wed, 30 Apr 2003 10:37:41 +0100

Hiya gurus
I'm suffering some horrible NFS problems at the moment.
Hardware is an E4500 running Solaris 8 (-12 kernel patch, upgraded to -20
last night, along with all prereq's for this patch level).
NIC is a ge2 sbus 1000sx card connected to a cat4003 switch.
Server is a NetApp Filer (This is not the problem, as another E4500 also
uses it as an NFS server, no problems at all - same NIC, same switch).
There are 25 mounts on the box, all with the following parameters:
/mf/TRAIN on 172.17.1.5:/vol/archmf/trainmf
remote/read/write/setuid/proto=tcp/rsize=16384/wsize=16384/intr/vers=3/hard/
dev=47c0012 on Wed Apr 30 10:13:47 2003

Every now and then (when the system seems to be under more load), NFS just
seems to freeze up completely. DF hangs after showing the local file
systems, console says "NFS server 172.17.1.5 not responding, still trying".
Again, this is unrelated to the load level on the filer. The system returns
to normal after several minutes of hanging.

netstat -i says:
ge0 1500 lonsun04.gb lonsun04.gb 736430 4434 580109 0 0 0
I know there shouldn't be any Ierrs here, but I figure a 0.5% error rate
shouldn't be a train smash of the order I'm experiencing.
netstat -s doesn't show any large numbers in any of the "worrying" fields.
nfsstat shows:
Client rpc:
Connection oriented:
calls badcalls badxids timeouts newcreds badverfs
164681 28 4 15 0 0
timers cantconn nomem interrupts
0 0 0 4

While the NFS is "hung" I can still ping the filer from the client, so it's
not like the switch has dropped the server or anything. The switch shows 0
input errors, and an autonegotiated 1000-full connection (as does the sun).
The NICs are running in local mac address mode:
hme0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
        inet 172.20.185.15 netmask ffff0000 broadcast 172.20.255.255
        ether 8:0:20:ab:78:2
ge0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 3
        inet 172.17.1.3 netmask ffff0000 broadcast 172.17.255.255
        ether 8:0:20:c5:fe:f7
so it's not mac address conflicts either.

Any ideas would be greatly appreciated

Thanks
Steven
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers

• Previous message: Bill Bradford: "ADMIN NOTE: Search engine back up"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]