Date: Thu, 04 Oct 2001 12:38:19 -0400 From: Mike Tancsa <mike@sentex.net> To: Luigi Rizzo <rizzo@aciri.org> Cc: freebsd-net@freebsd.org Subject: Re: How to increase TCP and UDP buffers (for IPv4/IPv6) Message-ID: <5.1.0.14.0.20011004122716.0408d470@marble.sentex.ca> In-Reply-To: <200110041628.f94GSXP45008@iguana.aciri.org> References: <21morto01so4jkebem9s5t1tf6ohm1uple@4ax.com>
next in thread | previous in thread | raw e-mail | index | archive | help
At 09:28 AM 10/4/01 -0700, Luigi Rizzo wrote:
> > > sysctl -w net.inet.ip.intr_queue_maxlen=100
> >
> > Hi,
> > Are there any nasty side effects for increasing this value ? Also, how
> > would one go about tracking down why net.inet.ip.intr_queue_drops is
> > incrementing ?
>
>In general, if your system is unable to drain ipintrq fast enough
>then you are hitting some CPU shortage, and increasing the queue
>size will not help, except perhaps in the short term or in peculiar
>situations where you have many interfaces, no fast_forwarding (this
>would help a lot here), or very bursty input traffic.
I guess in my case, the load average is general 0.00, but that is just
measuring userland activity no ? Is there a way to allocate more CPU to
the kernel for routing ? The box in question has some 100,000 routes, so I
dont think I can implement fast_forwarding-- or am I thinking of something
else?
Also, will removing ipv6 from the kernel make it more efficient acting as a
router ?
Thanks very much for your input.
---Mike
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5.1.0.14.0.20011004122716.0408d470>