Re: process creation time increases linearly with shmem

[Nick Piggin <nickpiggin@yahoo.com.au>]

Ray Fucillo wrote:

> Nick Piggin wrote:
>
>> How does the following look? (I changed the comment a bit). Andrew, 
>> please
>> apply if nobody objects.
>
>
> Nick, I applied this latest patch to a 2.6.12 kernel and found that it 
> does resolve the problem.  Prior to the patch on this machine, I was 
> seeing about 23ms spent in fork for ever 100MB of shared memory 
> segment.  After applying the patch, fork is taking about 1ms 
> regardless of the shared memory size.
>

Hi Ray,
That's good news. I think we should probably consider putting the patch in
2.6.14 or if not, then definitely 2.6.15.

Andrew, did you pick up the patch or should I resend to someone?

I think the fork latency alone is enough to justify inclusion... 
however, did
you actually see increased aggregate throughput of your database (or at 
least
not a _decreased_ throughput)?

> Many thanks to everyone for your help on this.
>

Well thank you very much for breaking the kernel and telling us about it! :)

Nick


Send instant messages to your online friends http://au.messenger.yahoo.com