Datalake
Runbooks
Operational playbooks for Datalake
Pub/Sub backlog growth
Symptoms
- Subscription backlog increases and
messages_processedstalls.
Checks
- Verify the subscription name and IAM permissions.
- Inspect Cloud Run CPU/memory usage and scale limits.
- Reduce
MAX_MESSAGESorMAX_BYTESif memory pressure occurs.
BigQuery insert errors
Symptoms
- Errors inserting into BigQuery; messages are nacked.
Checks
- Confirm dataset and
<tableName>_changelogtables exist. - Validate BigQuery permissions for the service account.
- Check schema compatibility with the changelog format.
Missing downstream topics
Symptoms
- Downstream consumers do not receive Pub/Sub updates.
Checks
- Verify topic creation permissions in the BigQuery project.
- Inspect logs for topic creation errors.
Health checks
GET /api/v1/healthreturns status and processing metrics.
Last updated on