Why hostname of docker embedded DNS does not work?

Question

I am trying to check if postgres container is accessible form within another container across the bridge docker network.

---

# /!\ Remove all containers /!\
docker rm -vf $(docker ps -aq)

---

Launch database

docker run --name    db                     \
           --env     POSTGRES_PASSWORD=asdf \
           --detach  postgres:alpine

---

Test connection with IP

IP=`docker network inspect bridge | grep IPv4 | grep -o -P '\d+\.\d+\.\d+\.\d+'`; \
docker run --env IP=$IP alpine sh -c "apk add --update --no-cache netcat-openbsd && nc -zv $IP 5432"

leads to:

Connection to 172.17.0.2 5432 port [tcp/*] succeeded!

Working!

---

Test connection with hostname

docker run alpine sh -c "apk add --update --no-cache netcat-openbsd && nc -zv db 5432"

leads to:

nc: getaddrinfo for host "db" port 5432: Name does not resolve

Not working...

Answer 1

You have two options:

1. Create a "link" between the two containers over the default bridge network.

docker run --name db --env POSTGRES_PASSWORD=asdf --detach postgres:alpine
docker run --link db alpine sh -c "apk add --update --no-cache netcat-openbsd && nc -zv db 5432"

2. Use a custom bridge network.

docker network create --driver bridge postgres-network
docker run --name db --env POSTGRES_PASSWORD=asdf --detach --network postgres-network postgres:alpine
docker run --network postgres-network alpine sh -c "apk add --update --no-cache netcat-openbsd && nc -zv db 5432"

Both solutions give:

Connection to db (192.168.128.2) 5432 port [tcp/postgresql] succeeded!

Note that the default bridge network is not recommended for production as explained in the Docker networking tutorial, so solution #2 is better.

If you use docker-compose instead of plain docker commands to run your containers, a custom bridge network is created automatically for each docker-compose.yml.