Question

A payments microservice on GKE writes structured JSON logs to Cloud Logging. The SRE team wants a Cloud Monitoring dashboard widget that shows the rate of failed transactions over time, where a failure is any log entry with jsonPayload.status equal to "DECLINED" or "TIMEOUT". They do not need to capture any numeric value from the log, only the frequency of these events. Which approach most efficiently produces the metric needed for this widget?

Accepted Answer

Create a counter-type log-based metric with a filter matching the two status values, then chart its rate on the dashboard.

Answer

Create a distribution-type log-based metric that extracts jsonPayload.status as the value field, then chart its mean.

Answer

Export all logs to BigQuery via a sink and query the failure count on a scheduled basis to populate the dashboard.

Answer

Enable Cloud Trace on the service and build the widget from span counts filtered by an error attribute.