Date: Sun, 3 Jun 2001 11:15:32 +0930 From: Greg Lehey <grog@FreeBSD.org> To: davep@afterswish.com Cc: freebsd-questions@freebsd.org Subject: Re: Hardware inturrupts mostly kiling 4.3R Message-ID: <20010603111532.R87716@wantadilla.lemis.com> In-Reply-To: <20010602035830.20776.cpmta@c001.snv.cp.net>; from davep@afterswish.com on Fri, Jun 01, 2001 at 08:58:30PM -0700 References: <20010602035830.20776.cpmta@c001.snv.cp.net>
next in thread | previous in thread | raw e-mail | index | archive | help
[Format recovered--see http://www.lemis.com/email/email-format.html] Your text was one line per paragraph. On Friday, 1 June 2001 at 20:58:30 -0700, davep@afterswish.com wrote: > The 4.3R machine that was acting as my gateway, webserver etc. was > working fine until yesterday. I moved a number of the machines it was > attached to, but didn't change the machine itself and now it's > behaving badly. > > Closer inspection reveals I'm getting around 290 hardware > interrupts/sec from *somewhere* and this is takng 75% of processor > time: > procs memory page disks faults cpu > r b w avm fre flt re pi po fr sr ad0 ac0 in sy cs us sy id > 0 0 0 36576 26236 11 0 0 0 7 0 0 0 293 109 21 1 76 23 > 0 0 0 36576 26236 5 0 0 0 0 0 0 0 286 18 7 0 75 25 > 0 0 0 36576 26236 3 0 0 0 0 0 4 0 290 22 8 0 75 25 > > > Apologies if that's illegible, That part was fine. It was the rest of the message which was mangled. > but it shows 290 interrupts/sec, approx 10 cs/sec, no dsk activity, > and page faults settle down to a constant 3/sec. Running ps ax shows > that no userland programs have any significant loadings. > > It's an old machine (P200) so there may be a hardware fault going > on. The PSU was making some interesting noises and once it failed to > power up the disk drive on boot, so it could be that on the way. The > disk is IDE and very new, so should be OK and fsck shows no errors > (is there a better check?). Could it be an error on the swap > partition? How would I check? It's unlikely to be a swap problem, since you don't have any disk activity, assuming you only have one disk. > So, is there any way I can find out where the interrupts are coming > from? I don't want to just bin the machine, or reinstall, because > that would be crap and I wouldn't learn anything. > > Please reply direct to this email address, like I said my 'main' > machine is elsewhere rght now and I'm having to do this off webmail > with a machine whose 'I' key s knackered! Hmm. This is an interesting problem. As Dan Nelson observed, 300 interrupts a second is nothing; when printing with a fast machine, it can go as high as 50,000. But the interrupt time is crippling: it seems that the interrupts are using too much time. In 4.3 there's no way to measure the time each individual interrupt handler is using (that will come in 5.0), but my guess is that this is the Ethernet card(s). Kernel profiling will tell you more, but it's a relatively complicated thing to do. Are you running a firewall? Are you possibly being hit by a lot of packets which fall all the way through the rules? Do a tcpdump on your external interface and see what you're getting. Greg -- When replying to this message, please copy the original recipients. If you don't, I may ignore the reply. For more information, see http://www.lemis.com/questions.html See complete headers for address and phone numbers 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?20010603111532.R87716>