Validators/Troubleshooting
/

Troubleshooting

502 Bad Gateway from the API

get_current_rounds failed: Server error '502 Bad Gateway'
for url 'https://api.neverplayalone.ai/validator/rounds/current'

What it means. A 502 comes from the proxy in front of api.neverplayalone.ai, not your validator. The backend was briefly unreachable — restarting, timing out, or under load.

What to do. Usually nothing. The validator logs a WARNING and retries on its next poll; a single 502 between healthy round_state lines is just noise. Only if 502s persist for many minutes is the backend actually down — then check GET /health, the API process, and its chain node. It's a server-side issue, not a validator fix.

The mission isn't switching

If the validator keeps logging mission=resource_gathering after you changed the mission:

  • The active mission comes from the roster, which is frozen at evaluation start. A round already being evaluated keeps its original mission.
  • Setting NPA_MISSION_ID on the validator only changes the fallback — the roster still wins. Change the mission on the backend, and it takes effect on the next round that freezes.
  • On a private test stack, delete and regenerate the roster to force the change immediately.

See Setting the mission.

The validator skips a round

If a validator starts more than NPA_EVALUATION_START_CUTOFF_RATIO (default 0.5) of the way into a round, it skips evaluating that round and waits for the next. This is by design — it avoids partial evaluations.

Docker / npabench errors

Evaluation needs a working Docker Engine and the vendored npabench. Re-run ./scripts/validator_setup.sh to reinstall dependencies, and confirm your user is in the docker group (the setup script adds it — you may need to log out and back in).