Date: Fri, 27 Aug 1999 01:04:48 +0200 From: Juergen Lock <nox@jelal.kn-bremen.de> To: Adrian Penisoara <ady@warpnet.ro> Cc: Alan Cox <alc@cs.rice.edu>, Juergen Lock <nox@jelal.kn-bremen.de>, scrappy@hub.org, freebsd-stable@FreeBSD.ORG Subject: Re: 3.2-STABLE hangs after several hours ... Message-ID: <19990827010447.A9022@saturn.kn-bremen.de> In-Reply-To: <Pine.BSF.4.10.9908261355110.5037-200000@ady.warpnet.ro>; from Adrian Penisoara on Thu, Aug 26, 1999 at 02:17:42PM %2B0300 References: <19990824153116.P39490@nonpc.cs.rice.edu> <Pine.BSF.4.10.9908261355110.5037-200000@ady.warpnet.ro>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Aug 26, 1999 at 02:17:42PM +0300, Adrian Penisoara wrote: > Hi, > > On Tue, 24 Aug 1999, Alan Cox wrote: > > > > > >In case anyone is looking at this, with as little info as there is here, > > > > >the last kernel updated was July 13th, so its between then and now that > > > > >the "bug" appears to have been introduced... > > > > > > > > I also went to update my kernel yesterday and stumbled across what > > > > appears to be the same problems as yours, hanging processes until the > > > > entire system becomes unusable... And here is what i came up with: > > > > > > It seems that I've been bitten by the same bug (the machine suddenly > > > freezes after some 2-3 hours); I'm checking out right now the submitted > > > patch and I'll let you know (probably tomorrow) how it works for me... > > > > > > > Please check if you have a process hung in "objtrm". Juergen did. > > I haven't the kernel debugger compiled in (as this is a production > machine); First, you don't need the kernel debugger to check a hung processes wchan, a ps -l will show it as well. (as does a ^t on its control tty, if it has one) > anyone care to point me out some documentatin suited to my > situation (kernel debugging on production machines) ? There's a section about kernel debugging in the handbook, http://www.freebsd.org/handbook/kerneldebug.html about production machines, well, you can still add ddb with the options DDB_UNATTENDED then the box will recover itself from a panic as it does without ddb. and the stuff about enabling crashdumps and building a kernel with -g applies to all machines anyway as long as you can spare the diskspace... > > > If so, this is the same bug reported in the "mSQL getting stuck > > in objtrm state" thread, and it has nothing to do with the changes > > below. (See my earlier message on this subject to the -STABLE list.) > > I checked out the previous threads and my situation appears to be similar > to those described in the "On freezes in 3.2-stable" thread, although I > have only 128Mb of RAM and UP kernel; see the attached dmesg output for > more details... > > Let me tell you how it worked for me: > > * I had an uptime of more than 13 hours with Juergen's patch, but that's > not so conclusive, as these freezes tend to be somewhat random > > * When I saw your commit I reverted the patch, cvsupped and rebuilt the > kernel; the machine hung up some 7-8 hours later ... > with the atomic.h fix (which alc mailed me) i cannot reproduce the hang anymore, at least the way i could with the broken kernels. so i would _guess_ that this hang you got had a different cause and it would be interesting to know what it was... > I really don't know what to blame, kernel bug or faulty hardware; one > thing I can be sure about is that the machine has been rock stable until > I've started to track -STABLE from 3.1-RELEASE somtime around 6th August. All -stable kernels after Jul 28 showed the hangs for me. Regards, -- Juergen Lock <nox.foo@jelal.kn-bremen.de> (remove dot foo from address to reply) 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?19990827010447.A9022>