Traffic from upnp is not sent to the shaper even when the ip address to which the upnp is pointing has a rule/queue associated with it.  I use pfsense miniupnpd.  I have one address on my network in the "penalty box" and when i enable upnp on utorrent for that address, it will go above my upperlimit.  When i turn off upnp on utorrent and use manual port forwarding the traffic is properly shaped.

This may explain why i was having issues with traffic shaping, multiple Xbox 360s and upnp…  if the upnp traffic is not subject to the traffic shaper then all the rules and queues in the world won't prioritize my Xbox 360 gaming traffic over my other traffic!  This is good to know.

Agreed.
1.2RC3 has some issues - at least some of the versions. But since we are lacking a build no. it's hard to tell which one…

Hi all,

could this be the problem? It seems it was fixed long ago, is this a regression of this problem?

http://forum.pfsense.org/index.php/topic,1246.0.html

Regards.

Maurilio.

Traffic shaping for download (WAN->LAN) traffic for Internet connections is fairly pointless in most cases. Really, all you can control is which traffic exits your LAN interface first which, for home/SOHO use, is not typically bandwidth constrained. Since you must accept traffic from your ISP in whatever order your ISP chooses to send it to you (FIFO), there isn't much, if any, benefit to be had. A caveat for pfSense is that you must have at least one WAN->LAN rule or the traffic shaper generator script breaks.

For upload (LAN->WAN) of course, traffic shaping is extremely useful.

Also, instead of rebooting the entire firewall, just reset states. Diagnostics, States, then click the Reset states tab. This will interrupt all established traffic flows, and when you reconnect they should fall into the right queues. You need to do this because applying a new traffic shaper policy doesn't affect connections that have already been established, only new connections.

i use aliases in firewall,for bandwidth shaping and penalty rules, and asign dhcp static for MAC

mm..  1 aliases : client1 ips: 192.168.100.50, 192.168.100.51
        2 aliases: cxlient2  1p: 192.168.100.52
then  trafic shaping penaltyBox
      adress:  client1
                  uP 128kb/s
                  Dw: 512kb/s
and for each aliases create penaltydown and penaltyuP queues and configure upperlimit m2  ex 256kb/s
this work fine for me
sorry my english

In http://forum.pfsense.org/index.php/topic,412.msg2559.html#msg2559 billm said:

ALTQ only shapes outbound on an interface, we create rules for BOTH interfaces and that's what the queues relate to.  An inbound (on the internal interface) and an outbound (on the external interface).

So why is there a direction field in the shaper rules? Isn't this always going to be 'out' ?

i am searching now to see if its possible to link pf statically…

the ordering needs to be PF before ipfw ?

ipfw can be linked statically quite easily...

edit: ok, dumb me :P pf can be linked statically...

im downloading pfsense developers edition and i will give it a try...

pf statically and ipfw as a module, lets see if this solves the problem...

edit2: while searching for this bug i found that theres not a lot of people trying to fix this bug, at least thats what i saw on the pf-freebsd list...

per IP, up pfsense 1,2

Hola wn.
I have a similar connection 4096KBps Down/512KBps Up and the best number so far is
3968 for qlanRoot , and 496 for qwanRoot
This is about 96.875% of connectioon speed  ie. 4096x0.96875=3968.
I,m using TERRA NETWORKS CHILE S.A.

also make sure you have values similar to these for the Traffic Shaper
34 %    Bandwidth qwanacks
  5 %    Bandwidth qlanacks
–-----------------------------------
Wait for someone with real knowledge to answer your question though.

Use this URL for your SPEED TESTS http://www.speedtest.net/index.php

No, unfortunately it is not possible.

Hi, I've been scouring the net for this specific problem, specifically that Remote Desktop (RDP) will saturate over a WAN connection when viewing highly graphical content, e.g. ppt files, graphically intense web pages, active content etc.

I was thinking about this, and realized that the only reason packet shaping works with downloads is because packets are dropped until the machine you are downloading from gets the hint to stop sending packets. Correct?

So, considering that in this situation you have one server with many RDP sessions, if you attempt to limit RDP downloads per session on the client side with packet shaping, this may not actually do anything because everything we want to limit is coming from the same IP. As far as the server goes, it will try and transfer X RDP packets, the server side internet connection will throttle packets (including printer stuff etc) down globally to match whatever the client is receiving, but most of what is transferred will still be whatever is hogging the queue.

I suspect that the optimal solution to this problem would be to stick the packet shaping box between the RDP server and that server's internet connection. That way you can truly limit where the bottlenecks are - packets sent from the server to the various WAN terminals. It would also be cheaper, since you'd only need one shaping box. After all, most of the upload at the WAN ends are just keyboard presses, mouse clicks and pointer positions over RDP, which should be fairly minimal and constant info. Will try it and see how we go.

Been exploring pfsense more (and tearing my hair out over captive portal, ill post about that later). It would appear that traffic shaping rules by schedule are not in pfsense so ill go and write some shell scripts instead. Consider this topic closed.