Subject: Re: syslogd restart / reboot on telnet?
To: CyberPeasant <djv@bedford.net>
From: George Coulouris <glc5@TC.Cornell.EDU>
List: port-i386
Date: 04/02/1998 13:20:27
> I think the full / filesystem is causing newsyslog to restart syslogd
> too often. WHen it cycles syslogd, I believe it may compress the old
> log files, thus freeing a little space on the disk. But inspect your
> /etc/newsyslog.conf to be certain about this, and "man newsyslog".

/ was only full for a short time; I was sorting some big files and filled
up /tmp. Then I setenv'd TMPDIR, which healed that problem; now / has a
comfortable amount of space.

> This doesn't look happy. Who's 101? Anything interesting?

That was me. See above :)

> > Apr  2 09:05:02 porter /netbsd: root on wd0a dumps on wd0b
> > Mar 31 21:55:27 porter savecore: /dev/wd0b: Device not configured
> 
> I see this too. Somebody else will tell us about it :) It should be
> trying to dump on wd1b, i.e. the swap partition. (Unless you're swapping
> on wd0b, too).

Nope, not swapping to wd0b.

> yeah:  clean up the wd0a partition, until you get many megs of free
> space on it. (10 MB free space makes me feel good).  I would look in
> particular (using find) for files owned by 101. Could it be a special
> user for running, say, an httpd server? That is logging a lot of things?
> If 101 is filling /, try symlinking its logfiles (or whatever) elsewhere.
> Maybe you have /home on /? symlink it away to somewhere else, like /usr/home
> if feasible.

Yep, /home is already symlinked to /usr/home.

> those lost interrupts are pesky. Maybe they'll go away. Maybe your disk
> or controller is going bad. Not many of those 40MB drives are still alive.
> (I have one too, on a headless 386SX). Check for loose cables. (Hope springs
> eternal in the hearts of men.) Bum power supplies can cause problems, too.
> (Been there on old machines.) Hmm, so can /heat/. You're at Cornell?
> It's been /hot/ in the E. US lately. (I'm in PA). The flakey memory in my
> old boxes starts acting up this time of year.

Yeah, this is the second 40M drive I've gone through. The first one died
about 20 minutes after I spun it up (granted, it hadn't been used in 10
years.. ). It might be a heat problem, things are packed pretty tight
inside.. I'll experiment with popping off some drive bay covers, more
fans, liquid helium, etc. and see what's what.