[ previous ] [ next ] [ threads ]
 
 From:  Michael <monowall at encambio dot com>
 To:  M0n0wall list <m0n0wall at lists dot m0n0 dot ch>
 Cc:  Anders HAGMAN <anders dot hagman at netplex dot se>
 Subject:  Re: [m0n0wall] Traffic shaper TOS and DSCP bits
 Date:  Sat, 17 Apr 2010 20:24:46 +0200
Hello Anders,

On Sat., Apr 17, 2010, Anders HAGMAN wrote:
> On 2010-04-16 12:03, Michael wrote:
>> I'm using M0n0wall 1.31 on PC Engines hardware.
>>
>> Referring to the page '/firewall_shaper_edit.php', there are
>> several choices for 'IP Type of Service':
>>
>>    lowdelay
>>    throughput
>>    reliability
>>    mincost
>>    congestion
>>
> The old eight bit TOS field looks like this.
> Tree precedence bits and five TOS bits.
>
> [ prec ][ D ][ T ][ R ][ C ][ Q ]
>     3     1    1    1    1    1
>
Okay, so now this is becoming clear. The first three bits relate
to precedence in the DSCP RFC. They are useful, but completely
ignored by M0n0wall's traffic shaper.

The last five bits which you labeled 'D, T, R, C, Q' refer to
the old IP4 TOS values 'lowdelay, throughput, reliability, mincost,
and congestion.'

If I am able to set these bits in packets arriving at the router,
then a M0n0wall traffic shaper rule can detect them according to
the radio buttons in the 'IP Type of Service' section at the
bottom of /firewall_shaper_edit.php.

CONCLUSION

It seems that that even when configuring traffic to use DSCP
CS5 (101000) as the QoS value, these packets will be ignored
by M0n0wall's traffic shaper unless matched by other means
(IP address or port.)

It would be nice if M0n0wall's traffic shaper could give priority
to packets marked with DSCP precedence CRITIC/ECP (critical -
emergency call processing.)

Because of this limitation I'll work around the problem by
reconfiguring all my high priority devices to send traffic using
the DSCP EF (expedited forwarding) value of 101110. It would seem
that EF's last three bits 110 will be matched by the traffic shaper
if I configure a rule setting the TOS values 'lowdelay' and
'throughput'.

Thanks for the explanation.

Regards,
Michael