Network Health

ssh root@your-server-ip

docker network ls

# Output looks like:
# NETWORK ID     NAME                DRIVER    SCOPE
# 2f3a8c8d3a01   bridge              bridge    local
# 8c91b1c39e10   host                host      local
# 4d2a91e0f4e2   none                null      local
# a1b2c3d4e5f6   myapp_default       bridge    local
# 7e8d9c0a1b2c   shared              bridge    local
# 9f0e1d2c3b4a   stale-old-network   bridge    local   <-- ghost? in use? who knows

docker network inspect shared --format '{{range $id, $c := .Containers}}{{$c.Name}}{{"\n"}}{{end}}'

# myapp-api-1
# myapp-redis-1

# Repeat for every network you might want to clean up.
# Match container names back to compose projects by string-matching prefixes.
# (Yes, that's the actual heuristic.)

# Cross-reference everything you found:
# 1. Network has zero containers? Probably ghost. Probably.
# 2. Network has containers from one project but you expected two? Membership broken.
# 3. Network appears in a compose file but not in 'docker network ls'? Missing.

# Now hope you got it right before you run:
docker network rm stale-old-network

# Re-run all of the above. Periodically.
# Or wait for production to break and figure it out then.