Path: utzoo!attcan!ncrcan!becker!censor!comspec!tvcent!lethe!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!rice!sun-spots-request
From: BKEHOE@WIDENER.BITNET
Newsgroups: comp.sys.sun
Subject: Have a little time?
Keywords: Miscellaneous
Message-ID: <7428@brazos.Rice.edu>
Date: 7 May 90 04:55:46 GMT
Sender: root@rice.edu
Organization: Sun-Spots
Lines: 57
Approved: Sun-Spots@rice.edu
X-Sun-Spots-Digest: Volume 9, Issue 150, message 16

I was just appointed Widener's "official" Sun systems manager, so now I
can finally ask a bunch of questions, since it's my responsibility to make
the fixes!  Here we go...hold on tight, folks. Please restrict any
flames/insults/guffaws to mail.

1. Yellow Pages. We've got one SS1 as the YP master, and another as its
   slave. To add an acct across the network, I do this:

   A. First, add the acct normally to /etc/passwd on the master.

   B. Now I go into /var/yp, rm passwd.time, and make passwd .. this will
      push the maps out containing the new acct.

So far I've had no problems with this part of it. It's when the user wants
to change their password that I get in trouble. I've got yppasswdd up with
the manual-suggested option of having it look at /var/yp/passwd for the
entries, to keep it from screwing with the master's files. But whenever
they do yppasswd, the change isn't ported properly. Could someone give me
the cut & dry way to properly set up YP so that it'll do everything it
*should*? The closest in TFM I could find was about using ypxfr to push
out changes in the maps -- but I didn't see it pushing out the passwd
changes at all.

2. Idle users. I'm sure there's something out there -- I need to watch for
   users that're idle for more than, say, an hour, and give 'em a couple of
   warnings before kicking them off (e.g. a problem with sunview left a user
   on ttyp4 for 3 days -- [just discovered this a little while ago]).

3. selection-svc. This process is periodically left after a person leaves
   sunview. It absolutely refuses to die properly & has to be manhandled. Any
   ideas? (This is just the beginning of my bitching about sunview..)

4. SunView. Where to start? We have a server up (SS1) with a diskless
   client. When anyone tries to use sunview on the client, it screams
   "Watchdog reset!"  at some point during sunview's initialization & the
   system has to be rebooted. Another error, "panic: Data fault" was solved
   via a patch tape from Sun. When they were asked about this, they said that
   it may even be a hardware problem (gee, can't read that in the manual can
   ya), but left it at that. Any- one have any practical experience with
   this? Everything works fine on the client otherwise (as far as using
   software goes).

5. hanging. While booting, the client sometimes sits for about 30-40 secs
   before kicking in, complaining that the NFS server isn't responding. We
   can't find any reason for this. Usually it takes a little activity on the
   server to kick the client into premature "birth" (e.g. doing netstat kicks
   it in for some reason).
6. logins. Is there a way to have a system-wide login file executed, in
   which I could add 'limit coredumpsize=0' to save on drive usage?
7. backups. Anybody have any sure-fire backup methods? My plan is to do a
   full backup every month, with incrementals nightly & weekly.

That's about it. If you're pulling out your hair at some (read: all) of my
questions, don't worry - yer in good company! Any help with these things
would be *greatly* appreciated.

-- Brendan Kehoe (bkehoe@widener.bitnet)