[linux-elitists] A better cron: large production environments

Asheesh Laroia linux-elitists at asheesh.org
Tue May 11 05:33:02 PDT 2010


On Mon, 10 May 2010, Don Marti wrote:

> begin Karsten M. Self quotation of Fri, May 07, 2010 at 07:57:22PM -0700:
>
>> This makes migrating numerous legacy scripts fairly transperant -- we're
>> adding one layer of infrastructure, but it's pretty light.
>
> Handling concurrency is getting to be important for
> things like this.
>
> If you have a multicore system and your cron jobs do
> a lot of network activity, it would be good to run a
> sensible number at once.  I have run into the dreaded
> cron job still running when it's time for it to run
> again problem.

For that, I suggest giving each cron job a named queue using "ts", a 
simple task spooler:

http://vicerveza.homeunix.net/~viric/soft/ts/

It will be in Debian "soon", it seems.

> (Is "load average" still even meaningful on a modern Linux system?)

I hope so! I hope I'm winning! 
http://www.asheesh.org/note/debian/load-average.html

(hmm, that page is loading slowly. I wonder what my load average is right 
now.)

-- Asheesh.

-- 
If you want me to be a good little bunny just dangle some carats in front
of my nose.
 		-- Lauren Bacall


More information about the linux-elitists mailing list