bregman-arie When to use
Use when services fail with quota exceeded, RESOURCE_EXHAUSTED, or similar quota limit errors.
Preconditions
- You can access the project and identify the failing service and region.
Procedure
- Extract the quota name/metric from the error message.
- Determine the scope: project, region, zone, or per-user.
- Confirm whether this is a spike (traffic/deploy) or a steady-state growth.
- Identify the top consumers (which resources or calls drive the quota).
- Choose mitigation: reduce usage, shift traffic, or request quota increase.
Decision points
- Spike caused by deploy/loop: rollback or hotfix to reduce calls.
- Legit growth: request quota increase and add alerting before hitting limits.
- Regional quota: spread across regions or rebalance.
Verification
- Error rate drops and quota errors stop.
- Resource creation/API calls succeed.
Rollback / undo
- Revert changes that increased consumption (deploy/config).
- If throttling is introduced, remove it after quota is safe.
Escalation
- Escalate to service owners if usage is driven by application behavior.
- Escalate to cloud owners for quota increase approvals.
Examples
Look for the quota metric name in the error, then map it to the service and region.