Troubleshooting

Symptom-first answers to problems hit during live deployment. Most startup issues come down to one of three things: a stale server still running, a stale .db file, or a slow network. See Running The Servers for normal startup.

Newly acquired data can’t be found in Tiled, even though the file was written to disk¶

A stale Tiled server is still running and serving an old save directory while new data is written somewhere else. Only one Tiled server should be active, bound to the current save path.

Find and kill the stale Tiled process (default port 9091), then restart it on the right path:

netstat -ano | findstr :9091
taskkill /PID <PID> /F

The cleaner fix is to let run_servers.py manage Tiled — it starts it as a tracked process bound to the save path you enter, and stops it on Ctrl+C. Changing data.save_path on the DATA device also restarts the managed Tiled server on the new path, so the server and the save path can’t drift apart.

Tiled (or the whole stack) fails to start after pulling new changes¶

Delete the stale .db files — both the Tiled catalog db and the Tango database db — then start fresh:

uv run startup_scripts/run_servers.py

A clean run rebuilds both databases.

Startup gives a weird error or an unexplained timeout¶

First suspect: old servers are still running. On the Windows microscope PC the PID changes constantly, so killing by PID is unreliable.

• Re-run run_servers.py and answer yes to Clear old processes first.

• To kill one specific Tango device server by name instead:

  import tango
  db = tango.Database()
  print(db.get_server_list("CORRECTOR/*"))
  d = tango.DeviceProxy("dserver/CORRECTOR/corrector_instance")
  d.command_inout("Kill")

Database won’t start, or “database did not become ready”¶

tango.Database() only connects to an existing database — it does not start one. The Tango database server must be started first (run_servers.py does this for you), or manually:

TANGO_HOST=localhost:9094 uv run python -m tango.databaseds.database 2

If the database server still won’t come up within ~2 minutes, enable tango.reset_database_file: true in the server YAML or check Delete tango_database.db before start in the startup GUI, then retry.

Servers stop responding (e.g. after repeated place_beam calls)¶

Reported recovery: kill the server notebook kernel, enable the Tango database file reset option, then start the servers again.

Unexplained timeouts when connecting from a remote location¶

Usually slow or unstable internet rather than a code fault. It has shown up alongside an unusably laggy Teams call, and once after a power flicker / thunderstorm — both times from a remote connection.

• Just retry — re-run the cell or command.

• Or remote into the microscope PC with TeamViewer and use VS Code there, so the connection to the servers stays local to that machine.

pandas version mismatch(observed - while using PyTemLib)¶

Recreate the environment:

# delete .venv, then:
uv sync