Summary

Design the API to support CLF to Stackdriver

Acceptance Criteria

• ClusterLogForwarder API changes approved by stakeholders
• API allows forwarding selected logs to different projects, regions, and/or log groups (a log group is similar to a Kafka topic)
• Document authorization requirements

Open questions

• Need to decide what to name the output type.
  • Most people say "stackdriver" but that is no longer the official name.
  • Was called Stackdriver, now GCP operations, Cloud Operations etc. Find official name.
• Does a mocking service exist that we can use for e2e or functional testing like cloudwatch

Notes

• Initially support to be for vector. We may forgo implementing in fluent entirely
• Follow pattern of Cloudwatch, Kafka and other output APIs:
  • https://github.com/openshift/cluster-logging-operator/blob/master/apis/logging/v1/output_types.go#L42

• Configuration:
  • Fluentd https://cloud.google.com/logging/docs/agent/logging/configuration#cloud-fluentd-config
  • Vector https://vector.dev/docs/reference/configuration/sinks/gcp_stackdriver_logs