Question

A microservices application on GKE writes logs to stdout as plain text lines. During incidents, engineers struggle to filter by severity and cannot correlate log entries belonging to the same user request across services. The team wants to improve log usability in Cloud Logging with minimal application rework and without deploying additional log-shipping agents. What should you recommend?

Accepted Answer

Emit logs as single-line JSON with fields such as severity and a trace/request identifier so Cloud Logging parses them into structured LogEntry fields for filtering and correlation.

Answer

Configure a log sink to export all container logs to BigQuery and have engineers write SQL queries to filter by severity and group by request.

Answer

Deploy a Fluentd DaemonSet with custom grok regex parsers to extract severity and request IDs from the plain-text log lines.

Answer

Increase the log retention period in the _Default bucket and rely on the Logs Explorer full-text search to locate related entries.