[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: machine temporarily unusable
- To: misc_(_at_)_openbsd_(_dot_)_org
- Subject: Re: machine temporarily unusable
- From: Jens Örtenholm <jens_(_dot_)_ortenholm_(_at_)_telia_(_dot_)_com>
- Date: Fri, 26 Oct 2001 01:26:09 +0200
> processes. This is not "expected" in a normal environment and suggests
> that either something went haywire and spawned zillions of processes, or
> due to a misconfiguration or some other error the process table was
> abnormally small.
I'm sorry, but I must've been thinking with some very inappropriate
part of my body. What struck me with surprise was that I, as an ordinary
user account, was able to (although by accident) push the system
to a point where I couldn't even do a root login.
Ofcourse, the mentioned zillion of processes were sendmail, and as
sendmail runs as root, any limits imposed on my user account (limits
of maxproc in login.conf comes to mind) doesn't really matter here.
I imagine sendmail would normally put up with this, but a faulty DNS
configuration made every spawned process think a bit too long, I
Would it, btw, be absolutely undoable to run sendmail as a user
`mail', or similar? It still bugs me a bit that such a simple
action made the whole system hit the roof.
> The ring buffer is written to disk on boot-up with just this sort of
> thing in mind. Look at /var/run/dmesg.boot for this. Quote inline
Ah, didn't know that one. Thanks.
Visit your host, monkey.org