Trouble with nfs

[Jocko]

Very very strange...

I am doing more tests (I reinstall again portmap).

A difference with your test: NWSP2 is the NFS Server and NWSP1 that is the client...

Edit
After 14 min the daemon "rpc.nsfd" is stopped !

Edit2
I tried a test where NWSP1 is the NFS Server. After 48min, the daemon "rpc.nsfd" is still running !!!

[fvdw]

ok I will start a trial with the other way around nwsp2 server and spd8020 client to see if that fails

[fvdw]

... the daemon "rpc.nsfd" is still running !!![/color]

How do you see that ?

in my process list it is shown as [nfsd] and not rpc.nfsd (also when using the top command)

[fvdw]

this is how it is listed in the process list on the nwsp2

Code: Select all

root@fvdwsl-nwsp2:/ # ps -ef
UID        PID  PPID  C STIME TTY          TIME CMD

1          991     1  0 18:26 ?        00:00:00 /usr/sbin/portmap
root       994     1  0 18:26 ?        00:00:00 /usr/sbin/rpc.mountd
root       996     1  0 18:26 ?        00:00:00 /usr/sbin/rpc.statd
root       999     2  0 18:26 ?        00:00:00 [lockd]
root      1000     2  0 18:26 ?        00:00:59 [nfsd]


this how it is listed on spd8020

Code: Select all

root@SPD8020:/ # ps -ef
UID        PID  PPID  C STIME TTY          TIME CMD

1        22248     1  0 18:23 ?        00:00:00 /usr/sbin/portmap
root     22251     1  0 18:23 ?        00:00:00 /usr/sbin/rpc.mountd
root     22253     1  0 18:23 ?        00:00:00 /usr/sbin/rpc.statd
root     22256     2  0 18:23 ?        00:00:00 [lockd]
root     22257     2 35 18:23 ?        00:53:05 [nfsd]



note that they are running already for almost 2.5 hours

[Jocko]

Yes, you are right : you must read "nfsd".

Bad news : after 1h, nfsd is no more working: CPU usage : 0%

My ps output

Code: Select all

NWSP1 :
1         2259     1  0 Dec03 ?        00:00:00 /usr/sbin/portmap
root      2262     1  0 Dec03 ?        00:00:00 /usr/sbin/rpc.mountd
root      2264     1  0 Dec03 ?        00:00:00 /usr/sbin/rpc.statd
root      2267     2  0 Dec03 ?        00:00:00 [lockd]
root      2268     2  0 Dec03 ?        01:24:24 [nfsd]

and top output for nfs
top - 20:59:49 up 19:48,  0 users,  load average: 0.03, 0.13, 0.53
Tasks:   1 total,   0 running,   1 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.3%us,  1.0%sy,  1.0%ni, 97.7%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:     12468k total,    11604k used,      864k free,      476k buffers
Swap:   257032k total,     8436k used,   248596k free,     3688k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 2268 root      20   0     0    0    0 S  0.0  0.0  84:24.70 nfsd


Output foor nwsp2

Code: Select all

 6365 ?        Ss     0:00 /etc/hostd 7200
25445 ?        Ss     0:00 /usr/sbin/portmap
25448 ?        Ss     0:00 /usr/sbin/rpc.mountd
25450 ?        Ss     0:00 /usr/sbin/rpc.statd
25738 ?        Ss     0:00 proftpd: (accepting connections)
25843 ?        D      9:01 /usr/sbin/rsync --stats --force --ignore-errors --modify-window=2 -HltDvpogr --delete /share/
25844 ?        S      0:00  \_ /usr/sbin/rsync --stats --force --ignore-errors --modify-window=2 -HltDvpogr --delete /sh
25845 ?        S     11:31      \_ /usr/sbin/rsync --stats --force --ignore-errors --modify-window=2 -HltDvpogr --delete


[fvdw]

Yes, you are right : you must read "nfsd".

Bad news : after 1h, nfsd is no more working: CPU usage : 0%

or it finished its job....mine is still running now in spd8020 as client and nwsp2 as server

[fvdw]

the backup failed when using sp8020 as client after less then 45 minutes.
After that the spd8020 had problems and needed a reboot to bring back to live.

probably this is because it lost the nfs connection with the nwsp2, then the processes using this connection will wait indefinetily on the connection to be restored and the nas hangs. Seems with mount options (soft instead of hard) you can prevent this waiting and after time out is passed the process gives up and nas becomes accesible again. Disadvantage will probably be unfinished backup..
I will reboot the spd8020 now and see if I can mount the nwsp2 again , on that one the nfs daemon is still running. If it mounts normally it must be a client issue would be my guess (eq mount)

[Jocko]

You don't need to reboot the NAs (NFS client).

Backup restarts and the NAS (NFS Client) runs well, if you do :
In the web interface of the NAS (NFS server), click on the button "start NFS server" in the NFS Menu ! :roll:

[fvdw]

I started some daemons in debug mode but cannot find a problem with nfs. All are still present when the process halts
But if you then try to access the nfs share via the mountpoint in putty then when you opne the mountpoint it just waits.
I mounted the nfs share for a second time on a new mountpoint and that also brought it back to live. So the problem seems to be in the mounting. Not yet clear what that problem is
ps trying to unmount the mountpoint as defined by the firmware is not possible as it is classified to be in use.

[fvdw]

I mounted the the nfs share now using tcp protocol instead of udp which is the default. tcp is better designed for flow control then udp.