From owner-freebsd-isp Sun Nov 15 18:04:21 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA01532 for freebsd-isp-outgoing; Sun, 15 Nov 1998 18:04:21 -0800 (PST) (envelope-from owner-freebsd-isp@FreeBSD.ORG) Received: from velvet.sensation.net.au (serial0-velvet.Brunswick.sensation.net.au [203.20.114.195]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA01457 for ; Sun, 15 Nov 1998 18:03:16 -0800 (PST) (envelope-from rowan@sensation.net.au) Received: from localhost (rowan@localhost) by velvet.sensation.net.au (8.8.8/8.8.8) with SMTP id MAA06620 for ; Mon, 16 Nov 1998 12:54:30 +1100 (EST) (envelope-from rowan@sensation.net.au) X-Authentication-Warning: velvet.sensation.net.au: rowan owned process doing -bs Date: Mon, 16 Nov 1998 12:54:26 +1100 (EST) From: Rowan Crowe To: freebsd-isp@FreeBSD.ORG Subject: Strange machine behaviour: sh / squid / pppd Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-isp@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Hello all, This isn't strictly an ISP specific problem, but I thought someone may be able to assist. Recently one of my machines (2.2.5-R, 6x86-120MHz, 64Mb RAM, 8.4GB HD) has started doing some very strange things... 1. sh was core dumping inconsistently on and off in a WWW crawler script that calls another sh script in a loop (I've _never_ seen sh core dump before!). This seems to have stopped now. 2. Idle time sometimes seems to be sitting at 0.0% for noticeable periods (as reported by top), with no apparent CPU hogs showing in the process display. top itself is usually the biggest "hog" at about 4%. There are currently 95 processes running. Here's part of top: last pid: 20335; load averages: 0.57, 0.98, 0.95 12:56:44 89 processes: 1 running, 87 sleeping, 1 zombie CPU states: 2.2% user, 0.0% nice, 7.0% system, 5.5% interrupt, 85.3% idle Mem: 25M Active, 6496K Inact, 20M Wired, 5544K Cache, 7681K Buf, 6052K Free Swap: 128M Total, 80M Used, 48M Free, 62% Inuse, 24K In PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 20334 root 34 0 644K 808K RUN 0:00 4.47% 2.14% top 8152 squid 2 0 32780K 11440K select 20:02 1.75% 1.75% squid 10175 root 4 0 1272K 780K bpf 14:18 0.65% 0.65% tcpdump 6019 root 2 0 356K 288K sbwait 50:39 0.04% 0.04% ipltd 10176 root -6 0 156K 200K piperd 5:30 0.04% 0.04% ip-mon.x 14071 root 2 0 316K 248K sbwait 28:10 0.00% 0.00% ipltd 3. Squid occasionally complains of an error via syslogd like: Nov 16 08:00:40 satin squid[19656]: xmalloc: Unable to allocate 4536 bytes! ...it then sits there for anything between 2-15 minutes (not accepting any connections during this time) before dying with a signal 6 and being restarted by RunCache. After that it works fine for another half day before dying again... I've pointed proxy.sensation.net.au to another backup machine for the moment but some customers will still be affected. 4. My ISDN PPP connection has been rudely severed a couple of times, but ip-down is *not* executed. I'm guessing pppd is dying suddenly. The interface is no longer marked UP or RUNNING but the source and destination addresses are still there. (I suspect gated may actually be the one doing the interface change perhaps after a timeout, it also correctly switches the default route to another link). The last 2 problems are Bad and Very Bad (in that order of importance 8-) ). I haven't yet rebooted the machine as I'm a little wary of doing it "just to see if it fixes things", especially on a production machine which is located remotely. Anyone have any ideas before I try rebooting when I can get on site? Thanks in advance.... Cheers. -- Rowan Crowe Sensation Internet Services, Melbourne Aust fidonet: 3:635/728 +61-3-9388-9260 http://www.rowan.sensation.net.au/ http://www.sensation.net.au/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message