automount, nfs server not responding

[jane zhang]

hi all,
our system is using nis and nfs automount, one of the sun box is very slow and I checked the messages file and it has:
Jan 9 16:51:31 dcssun46 automountd[671]: server stsja not responding

I asked the previous admin who set up the system and know stsja is not packed and retired, how do I find the nfs link between dcssun46 and stsja and get rid of this message?

dcssun46 is a sun box and stsja is a hp-unix box.

regards,
Jane

[Shannon Petry]

First, it is not really possible to give you a solid answer. NFS is very flexible especially when dealing with automount. Your NFS server can be defined in any map, so lets work on finding which one.

First, look at /etc/fstab (on Solaris /etc/vfstab) and ensure there is no mount specified to the non-responsive host.

Next, lets see what the auto_master map says to look at.

%ypcat -k auto.master

This should give entries back listing directories and map/file names.

Now you need to ypcat each of the maps listed to find the non-responsive host.

I.E. On one of my sites...
% ypcat -k auto.master
/net /etc/auto.home
/- /etc/auto.direct

If I do a more on /etc/auto.home
% cat /etc/auto.home
+auto.home

This means point to an NIS map of auto.home

%ypcat -k auto.home

joeshmoe somehost:/dir/joeshmoe

To simply find a hostname I can use grep after the command to look for what I need. In your cast dcssun46

%ypcat -k auto.home | grep dcssun46

Of course I get nothing, but you may.


I'd suggest if you are going to Administer UNIX and NFS systems, you get a book by O'Reilly and Associates called NFS and NIS Administration. Take a week of reading and experimenting to learn your way around a bit.


Regards,
Shannon

[jane zhang]

Shannon,
I already checked the items you mentioned in your email.

cat /etc/vfstab
It only shows the local disks of dcssun46.

dcssun46 has /etc/auto_home,/etc/auto_master, but the nis map name are auto.home, auto.master and auto.direct.
# ypcat -k auto.master
/- auto.direct -soft
/net -hosts -soft
/home auto.home -soft

# ypcat -k auto.home
* -rw &:/homes/&
# ypcat -k auto.direct | grep stsja return nothing.

# cat nsswitch.conf
automount: nis files

# cat /etc/mnttab | grep stsja return nothing.

I went through the nfs admin menual several times and could not find the answer and in our NIS domain, only this sun box is trying to mount stsja the down machine.

any other thoughts?

[Umesh Potbhare]

Jane,

There is possiblity that the active user on this machine has a path set to machine stsja in his .cshrc or .profile or check any file which is automatically getting sourced when the user is logging.

If this machine has entry into any of the paths also, it can cause this.

Get rid of that entry, reboot machine for safer side.

Also check what is the version of nfs on HP and what is the nfs version on dcssun46. If they are mismactching then also this problem might occure

Regards,
-Umesh

[Shannon Petry]

Sorry for the delay, not sure if your still having problems or not.

First, does the sun have a mount already to the down machine?

mount | grep stsja

If it reports a mount back, then you have to reboot the Sun to make the error go away, or get stsja back on line to unmount the bad directory.

The way that NFS works, a disconnect requires an acknowledgement from the server to complete. (lots of good reasons for this, so dont know why MS does not excepe they maybe expect to be down more).

If you look at the auto.home entry, basically this is a wildcard stating to mount everything from 1 spot to it's same name. This is probably not the cause.

It could be that someone manually made a mount for some reason, and never unmounted it. Thus, no entry in fstab, vfstab, auto.master.

Let me know the outcome.

Regards,
Shannon

[Kevin Wright]

ypcat -k auto.direct |grep stsja to see what FS is getting mapped to stsja.

OR
ypcat -k auto.home |grep stsja if it is your home directory server.

Is nfs running on stsja? rpcinfo -p stsja, or showmount -e stsja to check.

chances are you have a stale mount point, recycle NFS on stsja.

[jane zhang]

Hi all,
Thanks for the reponses, I worked with the user and find out there is one entry related to stsja in his library path in the design software.

So whenever the user try to copy some technology files, all the entries in the library path will be checked and since stsja is down and packed. The error message would appear in messages file.

It turns out that the nis and nfs automounter has bee working fine in the sun box.

Jane