Debug Backend Logs #

Use Grafana — it is the log path for developers. Open the gcp-operations or backend dashboard. It needs no GCP IAM beyond Cloudflare Access, covers full history, and can aggregate.

kubectl logs reads the same logs straight from one pod:

kubectl -n fs-apps logs deploy/backend

It shows the current pod’s output, so history ends at the last restart.

History across past pods #

Grafana holds the full history. kubectl logs reaches only the running pod, so anything from a previous pod needs Grafana or Cloud Logging.

Developers: use Grafana — the gcp-operations and backend dashboards wrap these queries in panels, need no GCP IAM, and already have access to the log bucket.

Counts, rates, grouping #

Cloud Logging cannot aggregate. It returns rows only, with no stats equivalent. Anything that counts or groups goes through the Grafana BigQuery datasource (funnelstory-bq-logs, both environments). Backend logs stream into a linked BigQuery dataset (backend_app_logs) for SQL.

SELECT
  TIMESTAMP_TRUNC(timestamp, MINUTE) AS t,
  JSON_VALUE(json_payload, '$.message') AS message,
  COUNT(*) AS n
FROM `PROJECT.backend_app_logs._AllLogs`
WHERE timestamp BETWEEN @start AND @end
GROUP BY t, message
ORDER BY t DESC

See Grafana for login steps.

Related #

• GKE diagnostics — find the pod, read why it restarted
• environment reference

Next: run the kubectl logs command at the top — it covers most day-to-day debugging.