Crash Recovery
This playbook describes how to recover safely after a process crash or host restart.
What Happened
- The process terminated unexpectedly.
- Orders or positions may have changed while the process was down.
Why It Happens
- OS restarts, power outages, or unhandled exceptions.
- Connectivity or auth failures causing process exits.
What To Check
logs/fyers_broker_errors.logandlogs/trading.log.- State DB health (
state_db_path) and file permissions. - REST connectivity and token validity.
What To Do Next
- Restart the process.
- Confirm the broker reconciles state on startup.
- Monitor for ReconcileFailureError and resolve before retrying.
Safety Notes
- Do not delete the state DB during live recovery unless you intend to reset order state.
- If the state DB is corrupted, stop trading and rebuild state only after manual verification.
Scenario Example
Scenario: The process restarts and fails with "state schema mismatch." What happened: - State DB schema version no longer matches the code. What to check: - Recent code changes or upgrades. What to do next: - Stop all processes, back up the state DB, then delete and restart to rebuild.