RS6K automounter is mounting me

[Brian Kelley]

The Automounter on the RS6K is rather disturbed.  IBM apparently didn't
like the way Sun and the rest of the industry had implemented the automounter
so they have enhanced it in a unique (to me) way.

On a Sun or DEC, if I cd to the automounted directory /turing and do a pwd, I get
the following:

euler>cd /turing
euler>pwd
/tmp_mnt/auto/turing


This is, of course, somewhat confusing, but livable.


On the RS6K, pain takes on a whole new meaning:

boat>cd /turing
boat>pwd
/tmp_mnt/auto51LwzMav

Real descriptive, eh?


An ls of /tmp_mnt reveals:

boat>ls -l /tmp_mnt
total 14
drwxr-xr-x   2 root     system       512 Jan 07 16:44 auto0Pu4CSM
drwxr-xr-x   2 root     system       512 Jan 08 21:01 auto150LglYbj
Drwxr-xr-x   4 root     staff        512 Dec 03 10:14 auto26LwxpW0
Drwxr-xr-x  29 root     usr         1024 Jan 21 16:29 auto33LwyEOL
Drwxr-xr-x   8 root     system       512 Jan 18 11:48 auto40LwyjYd
Drwxr-xr-x   8 root     system       512 Jan 14 23:49 auto47Lwy%a4
Drwxr-xr-x   6 root     staff        512 Dec 21 13:49 auto50LwzJZT
Drwxr-xr-x   7 root     staff        512 Jan 15 18:04 auto51LwzMav


Normally, the automounter dynamically creates links in this directory
with the same name as the remote filesystem you're accessing.  Obviously,
the RS6K is generating a sado-random name.  A good reason for this escapes me
at the moment.

One solution to this problem is to stop relying on vendors such as IBM to
provide a compatible solution and stick to something that works everywhere,
such as AMD.  This can get quite time-consuming.

IBM gets the code from Sun. You'd think they could spare the world from their
re-implementation for once.

- Is there a good reason for doing things the way IBM did?

- Obviously, totally breaking pwd like they did is not acceptable.
  Are there plans to correct this situation?

- If there are no plans to correct this, what do we need to do to make it
  happen?  (I hear it; switch to AMD...)



  Brian