Return to BSD News archive
Newsgroups: comp.unix.bsd Path: sserve!manuel.anu.edu.au!munnari.oz.au!bruce.cs.monash.edu.au!andrewh From: andrewh@cs.monash.edu.au (Andrew Herbert) Subject: Re: Occasional system hangs Message-ID: <andrewh.724059111@bruce.cs.monash.edu.au> Sender: news@bruce.cs.monash.edu.au (USENET News System) Organization: Computer Science, Monash University, Australia References: <1992Dec2.140837.12619@cli.di.unipi.it> <jason.723781864@sorokin> <ByysLH.Axr@ns1.nodak.edu> <1g95ueINNhri@manuel.anu.edu.au> Date: Fri, 11 Dec 1992 07:31:51 GMT Lines: 26 titus@coombs.anu.edu.au (titus chiu) writes: [386bsd hang description deleted] >In article <ByysLH.Axr@ns1.nodak.edu> tinguely@plains.NoDak.edu (Mark Tinguely) writes: >>this is the infamous VM problem. When swap is filled (or gets totally >>fragmented), the system can freeze. The nature of the VM makes this problem >>very difficult to totally solve. There are attempts to correct this problem >>for good. You can avoid the symptoms by adding more swap space. You must >hmmm .. we already have 24 megs of swap space and 8 megs of ram on the >system that jason described.. we could always add a 2nd swap i guess :P My system, with 16 MB RAM and 32 MB swap, hangs in a similar way. Existing processes such as INN processing incoming news and nntplinks sending outgoing news usually keep ticking away for quite some time (i.e. disk is unaffected), but it looks a whole lot like anything requiring a fork(), such as programs are run from the shell just hangs. Eventually all processes grind to a halt, but the system can still be pinged. The VM problem sounds quite a likely cause, alas. gcc 2.3.1 is particularly good at bringing about this problem, while X8514 and friends rarely lasts more than 12 hours without hanging 386bsd and requiring a reboot. Time for more RAM I guess! Andrew