nice(1) level during build

Declan Moriarty junk_mail at iol.ie
Mon Jul 3 07:28:59 PDT 2006


On Mon, 2006-07-03 at 15:55 +0200, Sebastian Faulborn wrote:
> >Hi. I just regained use of my system after 'make bootstrap' of gcc-4.1.1, with 
> >uClibc-0.9.28 and linux-2.6.17.1, when it consumed all resources for about 30 
> >minutes, and slowly gave resources back for 20 minutes after. I had a load 
> >average of 8 (new record), used all 1GB of ram, used all 1GB of swap, and all 
> >3ghz of cpu.
> >
> >I would like to suggest we start using 'nice -n 19 su - hlfs', and 'nice -n 
> >19 /tools/bin/bash' for chroot, unless anyone has objections.
> >
> >robert
> 
> If your system has a load of 8, uses all main memory and swap (!) just by running
> a single application (make bootstrap), then there must be something seriously wrong!
> 
> I have not tried gcc-4.1.1, but sofar all builds consumed at the most all main
> memory, no swap (at most 100kB), and a load of maybe 1.5.
> 
> So I rather prefered to fix the problem rather than setting nice levels.
> Normally a load of 8 signals a dead machine and is
> out of discussion for a production server (talking about single processor 
> machines). Allthough HLFS is not normally build on a production system - there
> is definitively something wrong!

I did run into something similar compiling some weird fpga program. I
ran out of memory and swap making a 200kb executable, with so many
includes and linked libraries that it needed over 150Megs to build.
That wasn't available, so it got killed by the kernel when I ran out of
swap, and reported

out of virtual memory (It was out of real memory too - I had free
running).

/# include no-expert-here warning
If it holds onto all memory and swap, I suspect something odd in the
code.
-- 
Declan Moriarty <junk_mail at iol.ie>




More information about the hlfs-dev mailing list