On Sun, Jun 29, 2003 at 08:50:52PM -0600, jared r r spiegel wrote:
> On Sat, Jun 28, 2003 at 11:37:06PM -0600, jared r r spiegel wrote:
> > 
> > pfctl: DIOCADDALTQ: Invalid argument.
> 
>   here's an odd one ( they're all odd to me <G> ).
>   i've now made two altq declarations.
>   one on the internal interface:
>  ( cbq, consisting of 118 queues total, parents and/or children )
>   and one on the external interface:
>  ( hfsc, consisting of 5 queues total, flat, no children outside
>    of the queues defined in the 'altq on' ).
> 
>   if the order they appear in the conf is:
> 
> altq cbq INT
> ( INT queues )
> 
> altq hfsc EXT
> ( EXT queues )
> 
>   i get the DIOCADDALTQ error.
> 
>   basically, any valid combination of order ( ie 'altq on' being 
>   above the queues which will be it's children ) in which the INT
>   ( cbq ) are above the EXT ( hfsc ) gives the error.
> 
>   if i put the EXT ( hfsc ) queues above the INT ( cbq ) queues in 
>   the pf.conf, i receive no error // parses/loads fine.
> 
> set require-order no
> 
>   made no difference.
> 
>   no big complaint, persay, it's something super easy to deal with
>   ( just put the hfsc guys on top, regardless of what interfaces
>     queueing is happening on, i guess ), so more or less i'm asking
>   in a round-about way if that is unexpected behaviour?

well yes, that's because the way the schedulers check for the # of 
queues is a bit stupid (just looking at the highest ID), and cbq 
allows for more queues than hfsc currently. hmm.

-- 
Henning Brauer, BS Web Services, http://bsws.de
[EMAIL PROTECTED] - [EMAIL PROTECTED]
Unix is very simple, but it takes a genius to understand the simplicity.
(Dennis Ritchie)

Reply via email to