IPFW + dummynet to lower NNTP traffic priority

Luigi Rizzo rizzo at icir.org
Wed Jul 27 09:43:58 GMT 2005


you must put a non-zero bandwidth on the pipe otherwise there is no
scheduling (0 means infinite bandwidth).
Also these are weights not priorities - even if one of the
queues has a very low weight it will still get some
bandwidth proportional to its weight.

cheers
luigi

On Tue, Jul 26, 2005 at 09:40:18PM -0600, Brad Waite wrote:
> I've been working on a way to keep my outbound NNTP traffic from
> interfering with all other traffic.  After reading the ipfw(8) man page
> and several other resources (including Luigi Rizzo's page), I thought I
> had it down:
> 
> ### Outbound
> ipfw pipe 1 config bw 0K
> ipfw queue 1 config weight 1 pipe 1
> ipfw queue 2 config weight 100 pipe 1
> ipfw add 1000 queue 1 all from any to any out dst-port 119 via fxp0
> ipfw add 1001 queue 2 all from any to any out via fxp0
> 
> ### Inbound
> ipfw pipe 2 config bw 0K
> ipfw queue 3 config weight 1 pipe 2
> ipfw queue 4 config weight 100 pipe 2
> ipfw add 1000 queue 3 all from [NNTP server IP] to any in via fxp0
> ipfw add 1001 queue 4 all from any to any in via fxp0
> 
> My thinking here is that both queues (in each direction) will share all
> available bandwidth, but any traffic heading out on port 119 or returning
> from the news serverwill be superseded by all other traffic.  If there's
> no other traffic, NNTP will be able to use all available bandwidth.
> 
> It would appear I'm not grasping the concepts.  Can anyone tell me what
> I'm doing wrong?
> 
> Thanks,
> 
> Brad Waite
> 
> _______________________________________________
> freebsd-ipfw at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-ipfw
> To unsubscribe, send any mail to "freebsd-ipfw-unsubscribe at freebsd.org"


More information about the freebsd-ipfw mailing list