[16340] in Athena Bugs
Re: sgi 8.2.9: /usr/tmp not cleaned
daemon@ATHENA.MIT.EDU (Greg Hudson)
Sun Sep 20 20:12:05 1998
To: Greg Hudson <ghudson@MIT.EDU>
Cc: "Bruce R. Lewis" <brlewis@MIT.EDU>, bugs@MIT.EDU
In-Reply-To: Your message of "Sun, 20 Sep 1998 19:46:31 EDT."
<199809202346.TAA01750@nephthys.grey17.org>
Date: Sun, 20 Sep 1998 20:11:57 EDT
From: Greg Hudson <ghudson@MIT.EDU>
Investigation revealed the following broken things:
* The files under /var/tmp didn't have recent access times.
Running /etc/athena/clean_tmp_areas nuked a whole bunch of
stuff, so clearly cron hasn't been running clean_tmp_areas.
* Output to /dev/console (including syslogs with appropriate
facilities/priorities) was not being displayed in the
console.
* In my root login shell, "logname" reported that I was
wschmitt, the last user to have a wtmp entry on ttyq3
besides root. "w" reports the correct entry for ttyq3
(root), and utmp and utmpx were both modified at 19:52 when
I logged in. logname and w both read the same file
(/var/adm/utmp). I think I understand what's going on
(login isn't overwriting wschmitt's DEAD_PROCESS utmp entry
and is instead writing a new one, probably because the ID it
chooses is different from what xterm chose, and logname
isn't skipping the earlier utmp entry), but I will have to
verify that theory and figure out what the right fix is.
For the first two cases, I don't know if this is just one singularly
broken machine or if lots of SGIs are behaving this way. I'll do some
testing tomorrow.
