[Pdns-users] rec_control with forking

[Nico van Royen]

_____  

From: bert hubert [mailto:bert.hubert at netherlabs.nl]
To: Nico van Royen [mailto:nico at van-royen.nl]
Cc: pdns-users at mailman.powerdns.com
Sent: Wed, 11 Oct 2006 21:26:29 +0200
Subject: Re: [Pdns-users] rec_control with forking

The 'fork' option is experimental. We have concrete plans to improve it, but
as it gives large financial benefits for users (you can basically run with
40% less servers), we are hoping someone is willing to sponsor 'enterprise
grade multi processor' support. 
You where pretty dead on with the 40%.. roughly 12k as opposed to 8k when the recursor is bound to both cores (dual core D805).
Using the --socket-pid with the rec_control now indeed gives some better statistics in the rrd graphs, thanks

Depends a bit. Both recursor and auth consume lots of memory *bandwidth*. It
appears there might even be a performance drop beyond 2G of memory use by
the recursor.
Any idea how much of a performance drop, or noticable in any case?
The authoritative server needs a lot less memory, and loves memory
bandwidth. 

In short -> get dual opterons.
Coming to that subject, and hoping not to start an intel/amd flame....  Does pdns itself have any liking to a particular hardware cpu (opterons, xeon, sparc, pa-risc etc?).  I read that SUN donated a T2000 for testing?..  kinda curious to the result on such a thing :)

Nico
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.powerdns.com/pipermail/pdns-users/attachments/20061013/0e980647/attachment-0001.html>