UNIX for Beginners Questions & Answers

Hello,

I got the below error on my AIX system when doing a df command

Code:

NFS server xxx not responding still trying

We check and know that the NFS server is not available anymore. So we would like to unmount it, but no help.

Code:

[root@yyyyyy] / > umount /mountpoint/
umount: Could not find anything to unmount

Also stop and start the nfs client, but still not help.
My AIX version is 6.1

--- Post updated at 09:02 AM ---

There is a way we can force the nfs to stop trying?

The problem is not the NFS itself but the mount process. But the very first question is: which NFS? NFS3 works completely different than NFS4 (and NFS2 is still - theoretically - available although only rarely used these days).

hmm... and it is still listed as "mounted" when you issue a mount command? Please check again because this looks unfamiliar.

How did you do that? NFS is controlled by the "system resource controller" and you have to stop/start it from there. FIrst, get the status (if you are unsure, please post the output of these comamnds, enclosed in CODE-tags):

Code:

lssrc -g nfs

Then stop it and/or start it. Before stopping it check for exported directories in /etc/mtab. If this file is not empty there are exports defined. Make sure other hosts are not affected by the stopping of the service.:

Code:

stopsrc -g nfs
startsrc -g nfs

Before restarting the NFS check if there are NFS mounts defined in /etc/filesystems.

Hi Bakunin,

Code:

How did you do that?

I use the method you mentioned, but use the parameter s, not g

Code:

stopsrc -s nfsd
startsrc -s nfsd

Thank you for your mount command. With mount command I found this:

Code:

[root@xxxx] / > mount
  node       mounted        mounted over    vfs       date        options
-------- ---------------  ---------------  ------ ------------ ---------------
         ...................................................
10.x.x.x /mount1   /mnt             nfs3   Apr 18 01:12

What I look in the config file /etc/filesystems:

Code:

/mount2:
        dev             = "/mount1"
        vfs             = nfs
        nodename        = 10.x.x.x
        mount           = true
        options         = intr
        account         = false

--> someone made the bad change here for some reasons.


Now I can force unmount the problematic mountpoint

Code:

[root@xxx] / > umount /mnt
umount: 16 error while unmounting 10.x.x.x:/mount1 - Device busy
[root@xxxx] / > umount -f /mnt
Warning: umount:: RPC: 1832-018 Port mapper failure - RPC: 1832-008 Timed out
forced unmount of /mnt

--- Post updated at 10:31 AM ---

Thank you for your info.
However, my problem is that someone made the bad change in configuration file, and it implies to my wrong action.

First a (minor) appeal: for quotes please use "QUOTE"-tags, not "CODE"-tags. The resulting formatting is different and it is easier that way to keep track of various parts of a posting. Thank you for your consideration.

The difference is that stopsrc -s only stops a single service (in this case nfsd) whereas stopsrc -g stops a group of services altogether. If you do a lssrc you will see the various groups the services belong to mentioned. You can use this group name to manage whole groups instead of single services. The same goes for startsrc and refresh.

NFS is not only nfsd but also portmapper, biod and statd (and maybe something else, i have no AIX system at hand and am quoting from failing memory). With the -g option you can manage the whole affected group (which is "nfs", not "nfsd"!).

LOL, i know the feeling. Have you ever considered using a sledgehammer on the persons private parts? This, i think, is called "best practice" as it not only drives home the message but also prevents procreation of such people. ;-))

I hope this helps.

bakunin