[Creator] [design-spec] gcp-cloud-loadbalancer-health

[rw-codebundle-agent]

Adds the gcp-cloud-loadbalancer-health CodeBundle, implementing design-spec issue #149.

Summary

This CodeBundle monitors the health and performance of GCP Cloud Load Balancers (external/internal HTTP/S, SSL proxy, TCP proxy, and network LBs). It discovers all forwarding rules in a project and analyzes each LB across four dimensions.

Tasks

• Discover GCP Cloud Load Balancers and Configurations — lists forwarding rules, categorizes by LB type, dumps config (IP, ports, target proxy, backend service)
• Check SSL Certificate Expiry — flags certs expiring within SSL_WARNING_DAYS on HTTPS/SSL LBs
• Check Load Balancer Backend Health — flags unhealthy/draining backends via backend-services get-health
• Analyze Error Rates — queries Cloud Monitoring 5xx ratios against ERROR_RATE_THRESHOLD
• Analyze Latency Performance — queries P95 latency against LATENCY_THRESHOLD_MS
• Generate Load Balancer Health Summary — aggregates all dimensions into a per-LB health summary table

SLI

Continuous 0-1 score averaged across four dimensions (ssl_certificate_health, backend_health, error_rate, latency_performance), each pushed as a sub-metric with the aggregate as the primary metric. Deployed via -sli.yaml template with - type: sli in generation rules.

Files

• runbook.robot, sli.robot
• 6 analysis bash scripts
• .runwhen/generation-rules + templates (slx/taskset/sli/slo)
• .test/ test infrastructure with Terraform fixtures covering healthy HTTP/HTTPS LBs, an expiring-cert HTTPS LB, and an unhealthy-backend LB

Deviations from spec

• Resource type is gcp_compute_forwarding_rules (full CloudQuery table name, matching the naming convention used by other GCP bundles).
• SLI backend_health dimension scored as healthy/total backend ratio per the spec.

[stewartshea]

I think this conflicts with a PR we already merged -- might have a bug in the back-end agent with a double submission