Date: Thu, 21 Mar 2002 02:20:26 -0500 From: Mike Nowlin <mike@argos.org> To: freebsd-stable@FreeBSD.ORG Subject: Re: 4.5-S crashing like clockwork Message-ID: <20020321022026.A17189@argos.org> In-Reply-To: <20020320022039.A2315@argos.org>; from mike@argos.org on Wed, Mar 20, 2002 at 02:20:39AM -0500 References: <20020320022039.A2315@argos.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 0, Mike Nowlin <mike@argos.org> wrote: > The one that I updated on March 7 (60G drive in it) decided to adopt a time > bomb mentality - it crashes like clockwork, literally, every 24 hours. > (Actually, it's usually 23:58:00 or somewhere around there - floats around a > minute or so - I'm guessing that might have something to do with the amount > of time the fscks take.) More information: (and some of good..:) ) After cvsupping and reinstalling last night, the system has been up for: akr-2-ctl-1:~$ uptime 2:08AM up 1 day, 1:15, 2 users, load averages: 0.00, 0.00, 0.00 akr-2-ctl-1:~$ ...and didn't blow up! I did make a couple little tweaks to the kernel config file: - added "-g" and am running off the debugging kernel - changed MAXUSERS=0 to MAXUSERS=32 (the 0 was from where I copied GENERIC and edited it) - uncommented the "device apm0" line (although it still has a "disable" tag in the line) I'm mildly interested in the MAXUSERS issue - could this have affected it? If so, why does GENERIC have 0 defined? (I'm guessing this isn't a problem.) ..and I'm REALLY interested in the uncommented apm0 line. After digging around in LINT, I found: # Notes on APM # The flags takes the following meaning for apm0: # 0x0020 Statclock is broken. # If apm is omitted, some systems require sysctl -w kern.timecounter.method=1 # for correct timekeeping. If this is a suspected (you tell me) trigger to causing these crashes, I'd be willing to play around with some experiments. (I'm going to try and set up another duplicate box tomorrow, set everything the way the crashing box was set, and see if it blows up - if so, I'll try the sysctl flag and see if that changes anything.) Comments? Questions? Ideas? (I'm wondering if the "for correct timekeeping" phrase may be a little understated. Maybe "for correct timekeeping, or if your box decides to crash every 24 hours." A little premature to make this suggestion officially, but it's something to think about....) I'll post my results soon. --mike To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020321022026.A17189>