pumba netem command with docker network not working

[sayankc]

Hi ,
My requirement is to create a docker container within a specific docker network and apply network chaos only on that network, so that other applications running on the host is not impacted.
Here is what I have done. But the delay is not working. only working with --interface eth0.

1. create a docker network.
   docker network create -d bridge --subnet 172.25.0.0/16 isolated_nw44

2. Run a docker container in this network.
   docker run -d -p 4444:4444 -v /dev/shm:/dev/shm --name=container44 --network=isolated_nw44 --ip=172.25.3.3 selenium/standalone-chrome

3. check the network interface created, where the docker container is.
   ip addr | grep 172.25 ==> inet 172.25.0.1/16 scope global br-72aff75d4744

4. Now applying netem on the new interface (br-72aff75d4744)
   ./pumba_linux_amd64.v0.6.5 netem --duration 5m --interface br-72aff75d4744 --tc-image gaiadocker/iproute2 delay --time 1000 container44

Result : It is not resulting any additional delay for the URL(http://ipadderss-of-host:4444/)

5. If I apply following command, netem is applying the delay but entire box network is impacted.
   ./pumba_linux_amd64.v0.6.5 netem --duration 5m --interface eth0 --tc-image gaiadocker/iproute2 delay --time 1000 container44

[alexei-led]

@sayankc I will check it.
Did you check with network interface assigned to your container (container44) binded to isolated_nw44 network?

[sayankc]

hi @alexei-led ,
I checked with isolated_nw44 too. Same result. network delay is not reflecting for the docker network.
./pumba_linux_amd64.v0.6.5 netem --duration 5m --interface isolated_nw44 --tc-image gaiadocker/iproute2 delay --time 1000 container44

[sayankc]

Hi @alexei-led ,
Could you please let me know if it is fixed / planned to be fixed in next release.

[alexei-led]

Hi @sayankc
Can you, please, post here output from ifconfig command running inside container44.

[alexei-led]

@sayankc does it work now? can I close the issue?

[sayankc]

@alexei-led , Issue is still there. I'll give you ifconfig.

[ebcFlagman]

I've the same issue.
Here's my docker-compose to reproduce:

version: '3'
services:
  web:
    image: nginx
  pinger:
    image: alpine
    command: "ping web"
  pumba:
    image: gaiaadm/pumba
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
    command: "--log-level debug netem --tc-image gaiadocker/iproute2 --duration 1m delay --time 5000 pumba-pinger-1"

[GruenerT]

Has there been a solution to this?

My own setup didn't work, but neither did the demo here:
https://github.com/alexei-led/pumba?tab=readme-ov-file#demo

My setup is running inside windows10 wsl2.

FROM ubuntu:latest

RUN apt-get update && \
  apt-get install -y iproute2 iputils-ping net-tools socat && \
  apt-get install -y bash kmod

CMD ["socat","tcp-listen:26379,reuseaddr,fork","tcp:redis-cache-layer:6379"]

then using
pumba netem --interface eth0 --duration 1h --tc-image gaiadocker/iproute2 delay --time 3000 latency-simulator

No delay is applied

[alexei-led]

Hi @GruenerT,

This is a known issue with how Pumba interacts with Docker networking, especially in WSL2 environments. A few things to try:

1. Check which network interface your container uses. Inside the container:

   docker exec latency-simulator ip route show default

   The interface name might not be eth0 — use whatever interface is shown.

2. Verify the tc rules are applied. While Pumba is running:

   docker run --rm --net container:latency-simulator ghcr.io/alexei-led/pumba-alpine-nettools tc qdisc show

3. WSL2 networking. WSL2 uses a NAT-based network which can interfere with tc netem rules. If the rules show as applied but have no effect, this is likely a WSL2 limitation.

4. Try the latest version (v0.12.3) with updated Docker SDK.

If tc qdisc show confirms the netem rules are present but traffic isn't affected, it's likely a WSL2/Docker Desktop networking issue rather than a Pumba bug.

---

🤖 Automated response by Marvin • alexei-led