Setting the file. One moment.
Chapter 103 · AWS Lambda Managed Instances
Subchapter 103.6
references/troubleshooting.mdMarkdown3 KBView on GitHub
| Issue | Cause | Resolution |
|---|---|---|
| 429 throttles during scale-up | Traffic doubled faster than 5-min scaling window | Increase MinExecutionEnvironments or lower TargetResourceUtilization |
| Function stuck in PENDING | Capacity provider provisioning instances | Wait several minutes; verify VPC subnets have IP capacity and IAM roles are correct |
| Architecture mismatch error | Function architecture ≠ capacity provider | Align both to arm64 or x86_64 |
| Cannot terminate EC2 instances | LMI instances managed by capacity provider | Delete capacity provider to destroy instances; cannot use EC2 console |
| High CPU, low throughput | Concurrency too high for CPU-bound work | Reduce PerExecutionEnvironmentMaxConcurrency to 1/vCPU |
| Race conditions in production | Code not thread-safe for multi-concurrency | Review with checklist in thread-safety.md |
| Function version not ACTIVE | Fewer than 3 execution environments ready | Wait for provisioning; check capacity provider status |
| Unexpected 500 errors | Unhandled concurrent access to shared state | Add thread-safe patterns from migration-patterns.md |
| CloudWatch logs missing | VPC egress not configured | Add NAT Gateway or CloudWatch Logs VPC endpoint |
| High costs despite low traffic | Minimum 3 instances always running | Evaluate if standard Lambda is more cost-effective |
Check capacity provider status: aws lambda get-capacity-provider --capacity-provider-name <name>
Verify subnets span 3+ AZs with available IPs
Confirm security group allows necessary egress
Check operator role has required permissions
Check for LMI-managed instances:
aws ec2 describe-instances --filters "Name=tag-key,Values=aws:lambda:capacity-provider" \
--query "Reservations[].Instances[].{Id:InstanceId,State:State.Name}"