Fix netem when destination IP filter is defined

[alexei-led]

Thanks to Alexey Krutilko

This need to be implemented inside Pumba netem command

The right sequence of commands to implement should be:

1. tc qdisc add dev eth0 root handle 1: prio - start new queue disc but no classes yet
   2.tc qdisc add dev eth0 parent 1:2 handle 30: netem delay 150ms - add 3 classes, 1:2 default class, all traffic goes through it with delay of 150ms
2. tc qdisc add dev eth0 parent 1:3 handle 20: netem delay 500ms - no new classes created, but delay is specified for 1:3, traffic still goes through 1:2
3. tc filter add dev eth0 parent 1: protocol ip prio 1 u32 match ip dst 172.17.0.3 flowid 1:3 - add destination filter by IP and now all traffic to this IP will go through class 1:3

[T-Eberle]

+1
Is this the reason why the IP filter doesn't work properly? A delay, for example, still impacts the traffic on all IPs.
Love the feature though!

[alexei-led]

Fixed in 0.4.7

[T-Eberle]

Works like a charm. Thx!