NFS4, iptables and evolution lockups

[John Austin]

Hi

Perhaps I should have known this
Perhaps I should be better at googling
but this took me a while to discover

Fully updated F13 client(milos) with users home directories NFS4
mounted on a fully updated Centos 5.5 NFS4 server (maui)

Evolution (in particular) falls over if the NFS4 call back port
has not been opened up on the client

Solution
Force the client to tell the server which port to use for call back
ja at milos ~ 1$ cat /etc/modprobe.d/nfs.conf 
options nfs callback_tcpport=801

Open up the port on the client
[root at milos ~]# cat /etc/sysconfig/iptables
...
-A INPUT -m state --state NEW -m tcp -p tcp --dport 801 -s "server_ip_addr" -j ACCEPT
...

I have no idea whether 801 is a good choice !

Symptoms

Following a clean boot of the client and only after login
of a user with home directory on the NFS4 server the server thows

maui kernel: nfs4_cb: server "client_ip_addr" not responding, timed out

No apparent problem until the user uses evolution
Then the directories "can" become unreadable and evolution is unusable

The server shows errors of the form
Oct 13 11:14:55 maui kernel: NFSD: preprocess_seqid_op: bad seqid (expected 2787, got 2788)
Oct 13 11:14:55 maui kernel: NFSD: preprocess_seqid_op: bad seqid (expected 2, got 3)
Oct 13 11:15:45 maui kernel: NFSD: preprocess_seqid_op: bad seqid (expected 2787, got 2788

Hope this is useful to somebody

John