[comp.sys.sgi] Amd5.3 ported to IRIX 3.3.1

srp@babar.mmwb.ucsf.edu (Scott R. Presnell) (09/25/90)

I had mentioned some time ago that I was trying to port amd (an alternative
automount daemon) to IRIX.  Some people had asked how that was going, I
don't have their addresses anymore.  Well, it's done, I think, for now.
I've only tried this on a 4D/20G running IRIX 3.3.1.

The difference was: 1) using amd5.3-alpha (from usc.edu in pub/amd).
and that my knowledge of irix has increased.

As a plug for amd - 5.3alpha has some pretty cool new features (new map
types) in it.  It's got alot to offer for special, varied mount situations.
Though for general purpose, SGI's port of automount seems just fine.

So if you want to, get the diffs from cgl.ucsf.edu (128.218.14.2) in
upload/amd5.3a.irix.shar. 

I'm still experimenting with the possible mount configurations at our site,
and I have not confirmed the operation of all the options.  Drop me a line
if you use it and find out something interesting. These diffs are also
going to the original author.

N.B.: Direct maps *DO NOT WORK!!* *DON'T* try them unless you want to lock
up your file system.  This also seems to be true for the IRIX 3.3
automount.  It doesn't seem to be an error in the programs, the pseudo
mounts seem to be set up correctly.  I'm curious: can anyone from SGI
comment on the problem?

	Thanks, and Good luck.

	- Scott Presnell (srp@cgl.ucsf.edu)

--
Scott Presnell				        +1 (415) 476-9890
Pharm. Chem., S-926				Internet: srp@cgl.ucsf.edu
University of California			UUCP: ...ucbvax!ucsfcgl!srp
San Francisco, CA. 94143-0446			Bitnet: srp@ucsfcgl.bitnet

sysjohn@physics.utoronto.ca (John Chee Wah) (09/27/90)

In article <srp.654278575@babar.mmwb.ucsf.edu> srp@babar.mmwb.ucsf.edu (Scott R. Presnell) writes:
# I had mentioned some time ago that I was trying to port amd (an alternative
# automount daemon) to IRIX.  Some people had asked how that was going, I
# don't have their addresses anymore.  Well, it's done, I think, for now.
# I've only tried this on a 4D/20G running IRIX 3.3.1.
# 
# The difference was: 1) using amd5.3-alpha (from usc.edu in pub/amd).
# and that my knowledge of irix has increased.
# 
# As a plug for amd - 5.3alpha has some pretty cool new features (new map
# types) in it.  It's got alot to offer for special, varied mount situations.
# Though for general purpose, SGI's port of automount seems just fine.
# 
# So if you want to, get the diffs from cgl.ucsf.edu (128.218.14.2) in
# upload/amd5.3a.irix.shar. 
# 
# I'm still experimenting with the possible mount configurations at our site,
# and I have not confirmed the operation of all the options.  Drop me a line
# if you use it and find out something interesting. These diffs are also
# going to the original author.
# 
# N.B.: Direct maps *DO NOT WORK!!* *DON'T* try them unless you want to lock
# up your file system.  This also seems to be true for the IRIX 3.3
# automount.  It doesn't seem to be an error in the programs, the pseudo
# mounts seem to be set up correctly.  I'm curious: can anyone from SGI
# comment on the problem?

3.3.1 automount (only tried direct map) does not work either. It does not
follow the symlinks it creates and the automounter cannot be restarted after
it is killed. I cannot remember if I reported it to the hotline under 3.3.
I too would like to hear what SGI has to say about this. Time to look at amd
again.

# 
# 	Thanks, and Good luck.
# 
# 	- Scott Presnell (srp@cgl.ucsf.edu)
# 
# --
# Scott Presnell				        +1 (415) 476-9890
# Pharm. Chem., S-926				Internet: srp@cgl.ucsf.edu
# University of California			UUCP: ...ucbvax!ucsfcgl!srp
# San Francisco, CA. 94143-0446			Bitnet: srp@ucsfcgl.bitnet

dana@tread.sgi.com (Dana Treadwell) (09/28/90)

Re: Direct maps

Right, the automounter in the IRIX 3.3.x release doesn't support the use
of direct maps. This would have required a major change to the IRIX file 
system code that we chose not to make. (We will be supporting direct maps
in a future release.) On the other hand, you can do everything via indirect 
maps (which has the advantage that it avoids "mount storms", where a poorly 
positioned 'ls' causes all of your automounts to happen).

Re: Killing your automounter

A "killed" automounter will act like a dead nfs server. If your automounter
is out-of-sync with the current contents of /etc/mtab, you can bring it up
to date by sending it a SIGHUP. If you really feel the need to kill your 
automounter, send it a SIGTERM so that it can die semi-gracefully.

    Dana Treadwell
    dana@sgi.com