[Pdns-users] Recursor 3.1.6-1 fork issue.

bert hubert bert.hubert at netherlabs.nl
Thu Oct 16 20:13:10 UTC 2008


On Thu, Oct 16, 2008 at 12:54:27PM -0700, Brad Dameron wrote:
> I know the forking was just a hack, and it does seem to improve
> performance. Is this just how this works and I should just let it be or
> is it not sharing the load?  

This is what many people see, and it is not a problem. UNIX operating
systems allow multiple processes to listen on a socket, but do not guarantee
any fairness in which process gets the packets.

It does however make sure that if one process is not taking packets, the
other one gets them.

So one of your cpus is busier than the other one, but this is not a problem
- packets get handled as soon as possible.

Perhaps you might want to swap your cores every once in a while in case one
feels it is having to work too hard :-)

	Bert

-- 
http://www.PowerDNS.com      Open source, database driven DNS Software 
http://netherlabs.nl              Open and Closed source services


More information about the Pdns-users mailing list