Date: Tue, 25 May 2004 14:57:40 +0300 From: Panagiotis Astithas <past@noc.ntua.gr> To: Ion-Mihai Tetcu <itetcu@apropo.ro> Cc: Konstantin 'Kosta' Welke <damaker@fillibach.de> Subject: Re: Freezes on heavy disk I/O Message-ID: <40B334B4.30101@noc.ntua.gr> In-Reply-To: <20040521194855.141f4757@it.buh.cameradicommercio.ro> References: <opr8cwfai3dvpa9u@auth.smtp.kundenserver.de> <40AE2776.30806@noc.ntua.gr> <20040521194855.141f4757@it.buh.cameradicommercio.ro>
next in thread | previous in thread | raw e-mail | index | archive | help
Ion-Mihai Tetcu wrote: > On Fri, 21 May 2004 18:59:50 +0300 > Panagiotis Astithas <past@noc.ntua.gr> wrote: > > >>Konstantin 'Kosta' Welke wrote: >> >>>Hi! >>> >>>I've got a box running FreeBSD 5.2.1. Sometimes, it just >>>freezes. Doesn't react to _any_thing except a hard reset. >>> >>>This is reproductable in some cases, all involving mainly >>>disk I/O. >>> >>>1) When burning a CD (on another computer) from a samba share >>>on the freebsd box, the box repliably freezes. This also >>>happens when I'm encoding a DVD rip to avi on my windows box, >>>while the DVD files are on the freebsd samba server. >>> >>>2) When doing stuff like portupgrade -a, it often (but always >>>at different files/ports) freezes. This happens especially >>>often when compiling gcc. (That's why I'm using the package now). >>> >>>3) Sometimes it just freezes like this, when playing an mp3, >>>surfing the 'net, etc. >>> >>>I used memtest enabling all test for 9 hours finding no errors. >>>Also, it's no overheating problem as it is not overclocked >>>nor is there any hot air coming out of the CPU fan when those >>>freezes occur. >>> >>>Any ideas? >> >>I have similar observations in my 5.2.1 machine. The recurring theme has >>always been that I was running under X and the freeze killed the >>console, although the machine responded to ping. >> >>When I tried the same action (a "portsdb -uU" or "portupgrade -a" in my >>case) from a virtual console I saw a panic. Since this never happens on >>my -current machine, I believe it is something already fixed in the tree. > > > Actually I think it still happens; I've had the need for a few hard > resets on my desktop with -current about 1 1/2 months; but it definitely > happen less that about 6 months or about a year ago, so it seems we're > on the right way :) I forgot to mention that putting "clear_tmp_enable=YES" in rc.conf makes it even more rare. It could be on of the softupdates bugs on full filesystems that got fixed (or almost:-)). Cheers, -- Panagiotis Astithas Electrical & Computer Engineer, PhD Network Management Center National Technical University of Athens, Greece
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?40B334B4.30101>