Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 23 Sep 1998 14:37:22 -0700 (PDT)
From:      David Wolfskill <dhw@whistle.com>
To:        freebsd-questions@FreeBSD.ORG
Subject:   inn-1.7.2 vs FreeBSD-2.2.6-R & resource limits?
Message-ID:  <199809232137.OAA07328@pau-amma.whistle.com>

next in thread | raw e-mail | index | archive | help
I have a dedicated 2.2.6-R machine for news.

I installed inn-1.7.2 from the ports collection as the news server.

It ran OK for a few months, and then I found out that I hadn't really
allocated the disk space properly (thanks in large part to UUNET's
unwillingness/inability/? to allow us to decline to receive articles
that are cross-posted to certain hierarchies -- at the time I was
setting things up, I had the impression that I'd be able to take
advantage of this).

I jury-rigged some circumventions (moving things around & replacing
the originals with symlinks to the real ones, mostly), and things
were quiet for a couple more months.

Then came a flood of newgroup control messages a few weeks ago (or so --
I lose track sometimes), and things have been hosed since.

I'm able to run innd as root OK, but if I start it up via the script in
/usr/local/etc/rc.d/innd.sh, it runs for a while, then dies (silently).

I haven't messed with the script; it looks like:

#!/bin/sh
if [ $# -eq 0 -o x$1 = xstart ]; then
   if [ -x /usr/local/etc/rc.news -a -f /usr/local/news/lib/history.pag ]; then
      limits -C news /usr/local/etc/rc.news && echo ' inn'
   fi
fi
if [ x$1 = xstop ]; then
        [ -x /usr/local/news/bin/ctlinnd shutdown machine is going down
fi


Now, the reason I quoted that is to show that it's using "limits".

And lately, I've also whacked the /etc/login.conf in an effort to
ensure that any limits associated with news on this system are
minimally constraining; here's the "news" stanza:

#
# Settings used by news subsystem
#
news:\
        :path=/usr/local/news/bin /bin /sbin /usr/bin /usr/sbin /usr/local/bin /
usr/local/sbin:\
	:cputime=infinity:\
	:filesize=infinity:\
	:datasize=infinity:\
	:stacksize=infinity:\
	:coredumpsize-cur=0:\
	:maxmemorysize=infinity:\
	:memorylocked=infinity:\
	:memoryuse=infinity:\
	:maxproc=infinity:\
	:openfiles=infinity:\
	:tc=default:

and here's the default one:

# Example defaults
# These settings are used by login(1) by default for classless users
# Note that entries like "cputime" set both "cputime-cur" and "cputime-max"

default:\
	:cputime=infinity:\
	:datasize-cur=22M:\
	:stacksize-cur=8M:\
	:memorylocked-cur=10M:\
	:memoryuse-cur=30M:\
	:filesize=infinity:\
	:coredumpsize=infinity:\
	:maxproc-cur=64:\
	:openfiles-cur=64:\
	:priority=0:\
	:requirehome@:\
	:umask=022:\
	:tc=auth-defaults:


And here's the "news" entry from /etc/master.passwd:

news:*:8:8:news:0:0:News Subsystem:/usr/local/news:/bin/csh

(for demonstrating that yes, the login class is set to "news").

But what I'm really wondering at this point is:  how can I tell what
resource is constraining innd -- assuming that I'm guessing at the cause
of the problem correctly?

I'm not seeing any messages in /var/log/messages (other than the usual
"internal rejecting huge article..." messages because I re-built inn
with a configuration parameter to drop any article bigger than 100KB on
the floor).

The only other thing I can think of that might be a little odd about the
machine is that I'm running ipfw on it -- I didn't see any point in
having the machine be aware of, let alone sensitive to, anything other
than NNTP and DNS from the outside world.

I welcome clues; I'm not (yet) too familiar with this rather new-fangled
login.conf stuff....  :-}

Thanks,
david
-- 
David Wolfskill		UNIX System Administrator
dhw@whistle.com		voice: (650) 577-7158	pager: (650) 371-4621

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199809232137.OAA07328>