Skip to main content

Fixed: "[Errno 48] Address already in use" When You Start the MLflow Server

· 7 min read
Jagdish Kumawat
Jagdish Kumawat
Founder @ Dewiride

uv run mlflow server --port 8080 stopped right after it began, with ERROR: [Errno 48] Address already in use and Running the mlflow server failed: The server process exited with code 3. Another MLflow server was already listening on that port. lsof showed which one, and the new server started cleanly on a free port.

What I Was Doing​

The tracking server for the MLflow series was running in one terminal, on port 8080. In a second terminal, in a different folder, I started another MLflow 3.18.0 server with the same command:

Terminal 2
uv run mlflow server --port 8080

What I Expected​

A second server starting, with Uvicorn running on http://127.0.0.1:8080.

What I Got​

Output
Backend store URI not provided. Using sqlite:///mlflow.db
Registry store URI not provided. Using backend store URI.
[MLflow] Security middleware enabled with default settings (localhost-only). To allow connections from other hosts, use --host 0.0.0.0 and configure --allowed-hosts and --cors-allowed-origins.
2026/11/13 17:47:12 INFO mlflow.server.jobs.executor_registry: Available backends after discovery: docker, local
2026/11/13 17:47:12 INFO mlflow.server.jobs.executor_registry: Job executor backend 'local' passed requirements check
2026/11/13 17:47:12 ERROR:    [Errno 48] Address already in use
Running the mlflow server failed: The server process exited with code 3. Please see the logs above for details.

A terminal where uv run mlflow server --port 8080 prints its first startup lines, then ERROR: Errno 48 Address already in use, and Running the mlflow server failed: The server process exited with code 3.

The command ended there and returned to the prompt. The first server kept running.

What I Checked​

What already listens on port 8080. lsof -nP -iTCP:8080 -sTCP:LISTEN lists the programs waiting for connections on that port:

Terminal 2
lsof -nP -iTCP:8080 -sTCP:LISTEN
Output
COMMAND   PID USER   FD   TYPE             DEVICE SIZE/OFF NODE NAME
Python  93838 you    3u  IPv4 0x40ce625db1a2fed3      0t0  TCP 127.0.0.1:8080 (LISTEN)
Python  93841 you    3u  IPv4 0x40ce625db1a2fed3      0t0  TCP 127.0.0.1:8080 (LISTEN)
Python  93842 you    3u  IPv4 0x40ce625db1a2fed3      0t0  TCP 127.0.0.1:8080 (LISTEN)
Python  93843 you    3u  IPv4 0x40ce625db1a2fed3      0t0  TCP 127.0.0.1:8080 (LISTEN)
Python  93844 you    3u  IPv4 0x40ce625db1a2fed3      0t0  TCP 127.0.0.1:8080 (LISTEN)

Five Python processes on 127.0.0.1:8080: an MLflow server's main process and its four workers, the one from the first terminal. How to Change the MLflow Server Port shows the same check on Windows.

That it's MLflow. It answered as one:

Terminal 2
curl -i http://127.0.0.1:8080/version
Output
HTTP/1.1 200 OK
date: Fri, 13 Nov 2026 13:36:00 GMT
server: uvicorn
content-type: text/html; charset=utf-8
content-length: 6
x-content-type-options: nosniff
x-frame-options: SAMEORIGIN
vary: Origin

3.18.0

Where the error comes from. MLflow's server runs on uvicorn, a Python web server. In uvicorn 0.54.0's uvicorn/config.py, when binding the address fails, uvicorn logs the error and exits with STARTUP_FAILURE, which is 3:

uvicorn/config.py (uvicorn 0.54.0)
            try:
                sock.bind((self.host, self.port))
            except OSError as exc:  # pragma: full coverage
                logger.error(exc)
                sys.exit(STARTUP_FAILURE)

MLflow's command line, mlflow/cli/__init__.py, then adds its own line: Running the mlflow server failed: … Please see the logs above for details. [Errno 48] is macOS's error number for an address that's already in use.

Why It Happens​

Only one program at a time can listen on the same address and port. The first server held 127.0.0.1:8080, so the second one's bind failed and it exited before starting any workers. Starting it from another folder doesn't help: ports belong to the computer, not to a folder.

The Fix​

Decide which of the two you want:

  • Use the server that's running. Usually that's the point: open http://127.0.0.1:8080 and carry on.
  • Replace it. Stop the first server with Ctrl+C in its terminal, then start the new one on 8080.
  • Run both. Give the second server its own port:
Terminal 2
uv run mlflow server --port 8081

Each server keeps its runs in the folder it starts from, so runs logged to one never appear in the other's UI. Point your code at the right one with its tracking URI: How to Set the MLflow Tracking URI in Python.

After the Fix​

The second server started on port 8081 and answered requests:

Output
Backend store URI not provided. Using sqlite:///mlflow.db
Registry store URI not provided. Using backend store URI.
[MLflow] Security middleware enabled with default settings (localhost-only). To allow connections from other hosts, use --host 0.0.0.0 and configure --allowed-hosts and --cors-allowed-origins.
2026/11/13 17:47:53 INFO mlflow.server.jobs.executor_registry: Available backends after discovery: docker, local
2026/11/13 17:47:53 INFO mlflow.server.jobs.executor_registry: Job executor backend 'local' passed requirements check
2026/11/13 17:47:53 INFO:     Uvicorn running on http://127.0.0.1:8081 (Press CTRL+C to quit)
2026/11/13 17:47:53 INFO:     Started parent process [19267]
...
2026/11/13 17:47:56 INFO:     Application startup complete.
2026/11/13 17:47:58 INFO:     127.0.0.1:55803 - "GET /version HTTP/1.1" 200 OK

lsof for both ports then showed a server on each: the first one, 93838, still on 127.0.0.1:8080, and the new one, 19267, on 127.0.0.1:8081.

Common Mistakes​

MistakeWhat happensDo this instead
Starting a "fresh" server without stopping the old one[Errno 48] Address already in useCheck the port with lsof first
Running two servers and expecting the same runs in bothEach has its own mlflow.dbUse one server, or log to the one you'll look at
Force-killing whatever holds the portYou may stop a server you, or another program, still needLook at what it is first, and stop it from its own terminal

Troubleshooting​

SymptomLikely causeFix
[Errno 48] Address already in useSomething listens on that port alreadylsof -nP -iTCP:<port> -sTCP:LISTEN, then pick a free port
The new server starts, but your runs are missingYour code logs to the other serverCheck mlflow.get_tracking_uri()

FAQ​

Why exit code 3?​

That's uvicorn's code for a failure while starting up, STARTUP_FAILURE. MLflow passes it on in its message.

Is [Errno 48] the same on Windows and Linux?​

The words Address already in use are what to look for; the number is macOS's.

Conclusion​

ERROR: [Errno 48] Address already in use followed by The server process exited with code 3 means another program, here a running MLflow server, already listened on the port. lsof -nP -iTCP:8080 -sTCP:LISTEN found it, and uv run mlflow server --port 8081 started the second server next to it.

Additional Resources​