Hi all,

I'm trying to debug a weird problem with nfsd on a 2.6.16.27-0.6-smp
kernel.

1 server: SuSE SLES 10 x86_64, config attached
256 clients: RHEL4 Update 4 2.6.9-42.ELsmp x86_64

Using nfs v3.

The clients have been happily talking to the server for several days
without incident.

The weird thing is that at a certain point the socket opened on port
2049 on the NFS server is being closed for unknown reasons (or better
for unknown reasons for me!). With unknown reasons I mean that I don't
see any critical error message in the logs, even with debug verbosity
enabled. I've enabled the max debug verbosity with:

echo 2147483647 > /proc/sys/sunrpc/nfsd_debug

The failures in the accept()s confirms that the server socket is working
fine and suddenly it has been closed (I did some attempts with a simple
netcat from localhost to check the socket availability):

# bzcat /var/log/messages-20071013.bz2 | grep accept
Oct 13 00:30:05 node0101 kernel: svc: tcp_accept ffff81015ea1d380 sock ffff81015ce1a780
Oct 13 00:30:05 node0101 kernel: svc: tcp_accept ffff8100c7997c00 allocated
Oct 13 00:30:06 node0101 kernel: svc: tcp_accept ffff81015ea1d380 sock ffff81015ce1a780
Oct 13 00:30:06 node0101 kernel: svc: tcp_accept ffff8100c7997c00 allocated
Oct 13 00:32:30 node0101 kernel: svc: tcp_accept ffff81015ea1d380 sock ffff81015ce1a780
Oct 13 00:32:30 node0101 kernel: svc: tcp_accept ffff8100c7997c00 allocated
Oct 13 00:32:31 node0101 kernel: svc: tcp_accept ffff81015ea1d380 sock ffff81015ce1a780
Oct 13 00:32:31 node0101 kernel: svc: tcp_accept ffff8100c7997c00 allocated
Oct 13 05:20:56 node0101 kernel: nfsd_acceptable failed at ffff8100c7873700
Oct 13 05:51:06 node0101 kernel: nfsd_acceptable failed at ffff8100cba472f0
Oct 13 09:51:11 node0101 kernel: nfsd_acceptable failed at ffff8100c6dbba40
Oct 13 11:04:33 node0101 kernel: nfsd_acceptable failed at ffff8100ce485a40
Oct 13 11:51:30 node0101 kernel: nfsd_acceptable failed at ffff8100c95552f0

The other strange thing is that the server receives a lot of requests to
close the connection from the clients, for example:

node0101:~ # bzcat /var/log/messages-20071013.bz2 | grep "close 1$"
...
Oct 12 18:07:15 node0101 kernel: svc: tcp_recv ffff810114f16dc0 data 1 conn 0 close 1
Oct 12 18:07:54 node0101 kernel: svc: tcp_recv ffff810114e07c80 data 0 conn 0 close 1
Oct 12 18:10:58 node0101 kernel: svc: tcp_recv ffff810114fcc880 data 1 conn 0 close 1
Oct 12 18:11:54 node0101 kernel: svc: tcp_recv ffff81010e940a80 data 0 conn 0 close 1
Oct 12 18:13:40 node0101 kernel: svc: tcp_recv ffff81013e4cd6c0 data 1 conn 0 close 1
Oct 12 18:13:45 node0101 kernel: svc: tcp_recv ffff810111acd8c0 data 1 conn 0 close 1
Oct 12 18:15:25 node0101 kernel: svc: tcp_recv ffff810112431c80 data 0 conn 0 close 1
Oct 12 18:16:00 node0101 kernel: svc: tcp_recv ffff810114f58980 data 1 conn 0 close 1
Oct 12 18:16:17 node0101 kernel: svc: tcp_recv ffff810114f58180 data 1 conn 0 close 1
Oct 12 18:16:27 node0101 kernel: svc: tcp_recv ffff81011130cc80 data 1 conn 0 close 1
Oct 12 18:16:37 node0101 kernel: svc: tcp_recv ffff81011130ca80 data 1 conn 0 close 1
Oct 12 18:17:03 node0101 kernel: svc: tcp_recv ffff81011130c880 data 1 conn 0 close 1
Oct 12 18:20:18 node0101 kernel: svc: tcp_recv ffff8100d67be9c0 data 1 conn 0 close 1
Oct 12 18:22:52 node0101 kernel: svc: tcp_recv ffff810111a23bc0 data 0 conn 0 close 1
...

Is it an expected behaviour or a potential symptom of a problem? Which
info could I search in the logs?

Any help appreciated.

Thanks,
-Andrea