Cloud Run

SmartSRE provides auto-scaling optimization and performance tuning for Google Cloud Run.

What SmartSRE Scans

Category	Checks
Performance	CPU/memory utilization, request latency, error rates
Scaling	Min/max instances, cold start frequency
Cost	Over-provisioning, idle resources
Security	IAM bindings, invoker permissions

Findings

High-Priority

Issue Type	Severity	Description
high_memory_usage	High	Memory > 90% for extended period
high_cpu_usage	High	CPU > 85% for extended period
high_error_rate	High	Error rate > 5%
oom_crashes	Critical	Out-of-memory terminations detected

Medium-Priority

Issue Type	Severity	Description
low_memory_usage	Medium	Memory < 20% utilization
cold_start_risk	Medium	Min instances = 0 with high traffic
high_request_latency	Medium	P95 latency > 2 seconds

Available Fixes

Memory & CPU Scaling

Operation	Description	Impact
scale_memory	Adjust memory allocation	Low
scale_cpu	Adjust CPU allocation	Low

Instance Management

Operation	Description	Impact
set_min_instances	Configure minimum warm instances	Low
set_max_instances	Set maximum scaling limit	Low

IAM & Security

Operation	Description	Impact
grant_invoker	Add service account as invoker	Medium
revoke_public_access	Remove allUsers invoker binding	High

Required Permissions

For Scanning

roles/run.viewer
roles/monitoring.viewer

For Remediation

roles/run.admin
roles/iam.serviceAccountUser

Example ChangeSet

{
  "service": "cloudrun",
  "intent": "Right-size memory allocation based on actual usage",
  "steps": [
    {
      "op": "scale_memory",
      "resource_ref": {
        "project_id": "my-project",
        "region": "us-central1",
        "service_name": "api-service"
      },
      "params": {
        "current_memory": "2Gi",
        "target_memory": "512Mi"
      },
      "estimated_cost_usd": -35.00,
      "impact_score": 15
    },
    {
      "op": "set_min_instances",
      "resource_ref": {
        "project_id": "my-project",
        "region": "us-central1",
        "service_name": "api-service"
      },
      "params": {
        "min_instances": 1
      },
      "estimated_cost_usd": 8.00,
      "impact_score": 10
    }
  ]
}

Configuration Options

Configure Cloud Run thresholds in Settings → Risk Policy:

Setting	Default	Description
max_memory_gi	8	Maximum memory allocation (GiB)
max_cpu_m	4000	Maximum CPU (millicores)
cpu_high_threshold	85	CPU % to trigger high_cpu finding
memory_high_threshold	90	Memory % to trigger high_memory finding
memory_low_threshold	20	Memory % to trigger over-provisioning finding

Rollback Capability

All Cloud Run operations support full rollback:

Operation	Rollback Method
scale_memory	Restore previous memory allocation
scale_cpu	Restore previous CPU allocation
set_min_instances	Restore previous min instances
set_max_instances	Restore previous max instances

Best Practices

1. Start with monitoring — Run scans before applying fixes to understand baseline
2. Right-size memory first — Memory affects billing and CPU allocation
3. Use min instances for latency-sensitive services — Prevents cold starts
4. Set max instances for cost control — Prevents runaway scaling