bind: Address already in use

1 A previous instance is still running (most common)

You stopped the terminal but not the process, or it forked into the background, or the crash left it alive.

$ sudo ss -tlnp | grep :3000

$ kill $(sudo lsof -t -i:3000)

Try plain kill first so the process can shut down cleanly. Only escalate to kill -9 if it ignores you.

2 Another service legitimately uses that port

Common collisions: 80 and 443 with an already-installed nginx or Apache, 3306 with MySQL, 5432 with PostgreSQL, 53 with systemd-resolved.

$ sudo ss -tlnp | grep -E ":(80|443) "

$ sudo systemctl stop nginx

Or change your program to a different port — often the better answer, since the other service is there for a reason.

3 TIME_WAIT after a restart

A socket that has just closed stays in TIME_WAIT for up to a couple of minutes. Restarting immediately can hit this, especially in test suites.

$ ss -tan | grep TIME-WAIT | head

$ # set SO_REUSEADDR in your server code, or wait ~60s

Most frameworks set SO_REUSEADDR by default. If yours does not, that is the correct fix rather than waiting.

4 You are binding to a privileged port without permission

Ports below 1024 need root. Some systems report this as an "in use" style failure rather than a permission error.

$ id && sudo ss -tlnp | grep :80

$ sudo setcap CAP_NET_BIND_SERVICE=+eip /path/to/binary

Cleaner than running the whole application as root: it grants only the ability to bind low ports.

5 A container is publishing the port

A forgotten Docker container holds the port from outside your usual process list.

$ docker ps --format "table {{.Names}}\t{{.Ports}}"

$ docker stop thecontainer