I am trying to throttle NNTP down to 256kbps per user/IP.

I think I am almost there, but now the clients show one nntp connection as very high and the rest as lower like they are being throttled.
This picture normally looks like 5-7 nntp connections all equal at about 3-400kbps (before the rules I entered).

I created the rule to mark the port 119 connection, then under it to mark the packet from the connection mark.
Then I created a PCQ queue type (left everything default) and then created a queue tree entry for it on the internal DMZ interface.

I just don't understand why it LOOKs like it's throttling all of their NNTP connection EXCEPT one that is now 4-5Mbps.

It's odd, but when I change the Queue it appears to throttle this connectino in torch back down, but then it climbs back up to what the picture looks like.

Almost as if his nntp client found a way 'around' the queue limit after a couple of seconds for that one connection (3-4Mbps).

paste the queue and queue type configuration from CLI.