Date: Mon, 6 May 1996 05:32:58 +1000 (EST) From: michael butler <imb@scgt.oz.au> To: gpalmer@FreeBSD.ORG (Gary Palmer) Cc: stable@FreeBSD.ORG Subject: Re: that was fun .. :-) Message-ID: <199605051932.FAA04040@asstdc.scgt.oz.au> In-Reply-To: <12520.831323817@palmer.demon.co.uk> from "Gary Palmer" at May 5, 96 08:16:57 pm
next in thread | previous in thread | raw e-mail | index | archive | help
Gary Palmer writes: > > May 6 03:42:51 asstdc innd: m:loadav [innwatch:load] 1792 > 1500 > > .. a load average of 1792 is just a tad more than I'd expect out of a > > 486DX4/100 :-) > Since that is innwatch, I THINK that that means the load is 1.792 (or > perhaps 17.92). It's a shell script, so gets rid of the decimal by > multiplying by either 100 or 1000... I'd be VERY impressed at a load > ave of 1792 for a FreeBSD box too :-) Ah .. <checks files> .. both the shell version and the perl one (which is what I run), edit out the '.', effectively multiplying by 100. The reason I made comment was that with only three streaming newsfeeds in full flight and a UUCP batching process that runs once an hour, it's near impossible to achieve load averages of 15 without it being an indicator of something being seriously wrong. Even a looping identd or innxmit only adds one to the load average. For whatever reason, not upgrading ld.so appeared to cause the UUCP batcher to spawn multiple shells all sitting there soaking up CPU without actually doing much useful or sensible. Replacing ld.so seemed to resolve it but I don't know why and sought to confirm that this might be an unexpected and undesired side-effect of the kernel and ld.so being out of synch, michael
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199605051932.FAA04040>