Docker "failed to create endpoint on network" in CI

By Daniel Zoghalchali·Latchkey

The daemon could not attach a container to a network. failed to create endpoint <name> on network <net> usually means a leftover endpoint/veth from a crashed container, an exhausted address pool, or a stale name still bound on the network.

What this error means

A docker run / docker network connect fails with Error response from daemon: failed to create endpoint <name> on network <net>: .... The network exists but the new endpoint cannot be created.

docker: Error response from daemon: failed to create endpoint quizzical_api on network bridge:
endpoint with name quizzical_api already exists in network bridge.

Common causes

A leftover endpoint from a crashed container

A container that died uncleanly can leave its endpoint/veth registered, so re-creating one with the same name collides.

The network address pool is exhausted

A bridge/network with too many attached endpoints (or a small subnet) can run out of addresses for a new endpoint.

Stale network state on the runner

Accumulated networks/endpoints from prior jobs leave the daemon network state inconsistent.

How to fix it

Disconnect the stale endpoint and prune networks

Force-remove the leftover container/endpoint and clean unused networks.

docker network disconnect -f bridge quizzical_api || true
docker rm -f quizzical_api || true
docker network prune -f

Recreate the network or use a fresh one

Remove and recreate a wedged network, or run on a clean per-job network.

docker network rm myapp_net || true
docker network create myapp_net

How to prevent it

• Clean up containers and networks between jobs on persistent runners.
• Use unique, per-job network/container names.
• Size custom network subnets for the number of endpoints you attach.

Frequently asked questions

Related guides

References

• Docker documentation
• Docker build cache
• GitHub Actions documentation