Our logging from the orchestrator and evaluators is not great.
- Consistently use msg or message or whatever is WMF-standard and adjust log dashboards according
- Determine standard
- Adjust code to standard
- Deploy image with new code
- Adjust dashboard columns/filters, if needed
- Have payload data actually show up, with the same things everywhere
- The x-request-id must be in every request-specific log message
- log API fetch boundaries with request IDs and timestamps
- evaluator: request received, response sent
- orchestrator: request received, response sent
- orchestrator: request sent to evaluator, response received from evaluator
- orchestrator: request sent to wikilambda_fetch, response received from wikilambda_fetch