docker: Bind for 0.0.0.0:PORT failed: port is already allocated

Hi, it's BlueByte. When Docker says a port is already allocated but docker ps looks empty, the culprit is usually a stopped container still holding the mapping. Let's look in the right place — including stopped containers — find the holder, and free the port cleanly.

What the error means

Starting a container fails:

Error response from daemon: driver failed programming external connectivity
on endpoint web: Bind for 0.0.0.0:8080 failed: port is already allocated

Docker cannot publish the host port because it is already taken — usually by another container it still tracks, occasionally by a plain host process. The confusing part is that docker ps can look empty while the port is still held, so the first step is knowing where to look.

Why the port is still held

A container that still holds the port mapping is bound to it. That container may be running, or — more confusingly — stopped but not removed after a crash or an interrupted docker run: Docker keeps reserving its port until the container is removed. That is why docker ps (running only) can look empty while the port is still "allocated". Less often, a non-Docker process on the host owns the port, or a Compose project left containers behind from a previous up.

Look for the holder — including stopped containers

docker ps hides stopped containers, so look for all of them:

docker ps -a --filter "publish=8080"

If a container is listed, that is your cause. If nothing is listed, the holder is not Docker — check the host directly:

ss -ltnp | grep :8080

That names the host process bound to the port. Between the two commands you will always know whether it is a container or a host process.

Free the port

1. Remove the container holding the port (stop it first if running):

docker rm -f <container_id>

2. If you use Compose, bring the project down so old containers are removed before recreating:

docker compose down && docker compose up -d

3. If a host process owns the port, stop it, or publish your container on a different host port:

docker run -p 8081:80 my-image

A real case: an exited container from yesterday

docker compose up fails with "Bind for 0.0.0.0:8080 failed: port is already allocated", but docker ps shows nothing running on 8080. You run docker ps -a --filter publish=8080 and find an exited container from a compose run that crashed yesterday — still holding the port. docker compose down removes it, docker compose up -d recreates cleanly, and the service binds 8080. The lesson: docker ps hid the culprit because it was stopped, not running.

Confirm the port was taken cleanly

Start the container again and confirm the mapping:

docker run -d -p 8080:80 my-image
docker ps --filter "publish=8080"

Seeing your container listed on 8080 means the port was freed and taken cleanly.

Keep it from recurring

Always docker compose down before up so old containers do not linger holding ports, and remove one-off docker run containers when you are done (or use --rm). On shared hosts, keep a note of which host ports each service publishes to avoid collisions.

How this differs from the "driver failed" chain error

port is already allocated is a conflict — the chain is fine, the port is just taken. That is the opposite of driver failed programming external connectivity ... No chain/target/match, where the port is free but Docker's iptables chain was wiped. Different cause, different fix — so check docker ps -a for a stopped holder first.