From owner-freebsd-hackers  Fri Jun 19 18:20:51 1998
Return-Path: <owner-freebsd-hackers@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id SAA16533
          for freebsd-hackers-outgoing; Fri, 19 Jun 1998 18:20:51 -0700 (PDT)
          (envelope-from owner-freebsd-hackers@FreeBSD.ORG)
Received: from MAIL1 ([195.216.60.4])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA16526
          for <freebsd-hackers@freebsd.org>; Fri, 19 Jun 1998 18:20:46 -0700 (PDT)
          (envelope-from daniel.sundin@engelholm.se)
Received: from p2.engelholm.se - 130.244.172.46 by engelholm.se with Microsoft SMTPSVC;
	 Sat, 20 Jun 1998 03:18:45 +0200
Message-Id: <2.2.32.19980620011743.00862a58@pop.engelholm.se>
X-Sender: 199706aZHerOgH@pop.engelholm.se
X-Mailer: Windows Eudora Pro Version 2.2 (32)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Sat, 20 Jun 1998 03:17:43 +0200
To: freebsd-hackers@FreeBSD.ORG
From: Daniel Sundin <daniel.sundin@engelholm.se>
Subject: state inode (top) problem
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

I'm having problems with a daemon made by myself. It's a daemon
that accepts connections and records the information of the 
person connecting to it, for the purpose of serving statistics
to the people using it. 

The daemon mmap()s a database file into memory, and reads/writes
from/to it there. It's been running fine for several months, but 
now suddenly it stops for 5-6 seconds at a time, accepting no connections 
at all. 

The database has slowly grown the last few months, and is now about 60mb in
size. The server has 256mb ram, and all the per-login/per-process memory
usage limits are set to infinity for the user the daemon runs under. The
FreeBSD version installed is a -current, freebsd 3.0 (980520-SNAP).

What happens when it pauses for 5-6 seconds, according to top, is that it 
goes into an 'inode' state. During that time the CPU usage and SIZE usage
decreases alot, but RES stays normal. 

mmap() should in no way flush changes to disk as often as this happens, 
or atleast not take aslong to write changes as it does. 

It repeats the 5-6 second inode state about once per minute, always when
the memory usage (SIZE) has grown to about 55mb (this differs a bit depending
on the memory currently available, the 'limit' was less when only 128mb ram
was installed). After the 5-6 seconds in inode state and a SIZE/CPU usage
that is way lower. When it gets out of the inode state both SIZE & CPU usage
starts to grow again. 

I'd appricate it alot if someone could either tell me what this
might be caused of, and/or what the STATE inode in top means exactly. 
The man pages for -current hasnt been updated to describe the new 
states.


Thanks in Advance,
Daniel



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