[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: race condition with reap_children when $(shell) used
From: |
Grant Taylor |
Subject: |
Re: race condition with reap_children when $(shell) used |
Date: |
Sun, 11 May 2003 15:33:33 -0400 |
> It is impossible to use the load average on its own to get good
> control over job spawning, you always wind up with a spike that
> brings a system to its knees. It was a bad idea from day one.
Of course, it's intended more as a "nice" feature for other users so
that when a system is busy make doesn't always run the -j number of
jobs.
>> IMHO it's entirely reasonable to specify something like -j8 -l4;
> When you're using tokens the load average specifier is completely
> superfluous. You're guaranteed that at most 8 jobs will be
> active. If you have 8 jobs running and the load average is less than
> 4, you won't get more jobs spawned to bring the load average up to
> 4. It's completely useless - harmful when used by itself, and
> pointless when tokens are in use.
No, it's entirely useful. The intent of -l is only to limit the
impact of a make on other users. It does in fact work well when used
this way; if two people run make -j8 -l4, their makes will tend to
back off so that the system won't crawl.
Anyway, regardless of what one is supposed to *use* these features
for, they've got bugs, and that's my main concern.
--
Grant Taylor - address@hidden - http://www.picante.com/~gtaylor/
Linux Printing Website and HOWTO: http://www.linuxprinting.org/