Robert Collins wrote:
> I don't know if this is on the to-do list, but I'm thinking of coding up
> self-adjusting helper counts.
>
> something along the lines of
>
> n is the number of running helper of a given type
> when new jobs are submitted, if the queue length for n is more than 2 * n,
> and no helpers have been spawned for n/2 jobs spawn n/10 (rounded up) new
> helpers.
> if the queue length stays at zero for n/2 jobs kill n/10 helpers.
>
> Yes I know magic numbers are bad - I'll have them in the conf..
>
> Comments/already done/bad idea?
>
> Rob
I'm just experimenting with skipping accept() if the queue length is over a
arbitrary number.
D
Received on Thu Sep 14 2000 - 17:08:55 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:37 MST