qpid-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andy Li <andr.deve...@gmail.com>
Subject Re: [Fwd: Re: Flow control behavior of fanout exchange]
Date Thu, 05 Nov 2009 00:31:10 GMT
We're running CentOS 5.3 on HP ProLiant DL360 G5. We haven't done any OS
tuning, such as disabling SMI's to lower the worst case latency. Just used
out-of-the-box config. (We haven't done any tuning on the "fast broker"
4-core machines either).

Why would lowering the # of working threads to #cores/2 change the behavior
of the broker so radically? Is there anything that can be done to let qpid
take advantuage of all the cpu cores?

Thanks,
Andy


On Wed, Nov 4, 2009 at 10:24 AM, Carl Trieloff <cctrieloff@redhat.com>wrote:

>
>
> Here is the deck,
>
> It may be a list of items on the hardware, it could be SMI's. Or it may be
> using the wrong timers for the hardware etc.
>
> Are you able to run RHEL5.3 or later (has more effective futexes) & give me
> the exact server model and I could probably tell you how to correct the
> issue.
>
> Carl.
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message