[intake] Monitor the health of LiteLLM on Kubernetes beyond contai...

[rw-codebundle-agent]

Request

Title: Monitor the health of LiteLLM on Kubernetes beyond contai...

Description:

Monitor the health of LiteLLM on Kubernetes beyond container logs -- API driven insights, request failures, budgets, and so on.

Platform: Kubernetes, LiteLLM

Additional context: port-forward for internal API access with kubectl

---

Existing Coverage (from registry search)

The following CodeBundles may overlap with this request. Designer: consider reusing, extending, or differentiating.

1. Azure ACR Health Check (95% match)

• Collection: rw-cli-codecollection
• Platform: Azure
• Description: Comprehensive health checks for Azure Container Registry (ACR), including network configuration, resource health, authentication, storage utilization, pull/push metrics, and security analysis.
• Tasks: Check Network Configuration for ACR ${ACR_NAME} In Resource Group ${AZ_RESOURCE_GROUP}, Check DNS & TLS Reachability for Registry ${ACR_NAME}, Check ACR Login & Authentication for Registry ${ACR_NAME}, Check ACR SKU and Usage Metrics for Registry ${ACR_NAME}, Check ACR Storage Utilization for Registry ${ACR_NAME}, Analyze ACR Pull/Push Success Ratio for Registry ${ACR_NAME}, Check ACR Repository Event Failures for Registry ${ACR_NAME}, Check ACR Security Configuration and RBAC for Registry ${ACR_NAME}
• Tags: ACR, STORAGE, REGISTRY, CONTAINER, NETWORK, HEALTH, AZURE, SECURITY
• Link: /collections/rw-cli-codecollection/codebundles/azure-acr-health

2. Kubernetes API Server Health (90% match)

• Collection: rw-public-codecollection
• Platform: None
• Description: Check the health of a Kubernetes API server using kubectl.
• Tags: KUBERNETES, AKS, EKS, OPENSHIFT, GKE
• Link: /collections/rw-public-codecollection/codebundles/k8s-kubectl-apiserverhealth

3. Kubernetes Redis Healthcheck (87% match)

• Collection: rw-cli-codecollection
• Platform: Kubernetes
• Description: This taskset collects information on your redis workload in your Kubernetes cluster and raises issues if any health checks fail.
• Tasks: Ping ${DEPLOYMENT_NAME} Redis Workload, Verify ${DEPLOYMENT_NAME} Redis Read Write Operation in Kubernetes
• Tags: KUBERNETES, REDIS, AKS, EKS, OPENSHIFT, GKE
• Link: /collections/rw-cli-codecollection/codebundles/k8s-redis-healthcheck

4. AWS CloudWatch Logs health (83% match)

• Collection: aws-c7n-codecollection
• Platform: AWS
• Description: Check AWS Monitoring Configuration Health
• Tasks: List CloudWatch Log Groups Without Retention Period in AWS Region ${AWS_REGION} in AWS Account ${AWS_ACCOUNT_NAME}, Check CloudTrail Configuration in AWS Region ${AWS_REGION} in AWS Account ${AWS_ACCOUNT_NAME}, Check for CloudTrail integration with CloudWatch Logs in AWS Region ${AWS_REGION} in AWS Account ${AWS_ACCOUNT_NAME}
• Tags: CLOUDWATCH, AWS, TAG, CLOUDCUSTODIAN, CLOUDTRAIL
• Link: /collections/aws-c7n-codecollection/codebundles/aws-c7n-monitoring-health

5. Kubernetes cert-manager Healthcheck (78% match)

• Collection: rw-cli-codecollection
• Platform: Kubernetes
• Description: Checks the overall health of certificates in a namespace that are managed by cert-manager.
• Tasks: Get Namespace Certificate Summary for Namespace ${NAMESPACE}, Find Unhealthy Certificates in Namespace ${NAMESPACE}, Find Failed Certificate Requests and Identify Issues for Namespace ${NAMESPACE}
• Tags: KUBERNETES, CERT-MANAGER, AKS, EKS, OPENSHIFT, GKE
• Link: /collections/rw-cli-codecollection/codebundles/k8s-certmanager-healthcheck

6. Kubernetes Cluster Node Health (75% match)

• Collection: rw-cli-codecollection
• Platform: Kubernetes
• Description: Evaluate cluster node health using kubectl
• Tasks: Check for Node Restarts in Cluster ${CONTEXT} within Interval ${RW_LOOKBACK_WINDOW}
• Tags: KUBERNETES, AKS, EKS, OPENSHIFT, GKE
• Link: /collections/rw-cli-codecollection/codebundles/k8s-cluster-node-health

7. Kubernetes Daemonset Health Check (71% match)

• Collection: rw-public-codecollection
• Platform: None
• Description: Checks that the current state of a daemonset is healthy and returns a score of either 1 (healthy) or 0 (unhealthy).
• Tags: KUBERNETES, AKS, EKS, K8S, OPENSHIFT, GKE
• Link: /collections/rw-public-codecollection/codebundles/k8s-daemonset-healthcheck

8. Kubernetes Patroni Health Check (67% match)

• Collection: rw-public-codecollection
• Platform: None
• Description: Uses kubectl (or equivalent) to query the state of a patroni cluster and determine if it's healthy.
• Tags: POSTGRESQL, KUBERNETES, AKS, EKS, PATRONI, OPENSHIFT, GKE
• Link: /collections/rw-public-codecollection/codebundles/k8s-patroni-healthcheck

---

Open Requests (may overlap)

Consider commenting on an existing issue instead of duplicating work.

• #82
• #64
• #48
• #86
• #85

---

Created via CodeCollection Registry intake at 2026-04-16 00:52 UTC.

[rw-codebundle-agent]

🔧 Creator agent starting (architect mode). Target: runwhen-contrib/rw-cli-codecollection

[rw-codebundle-agent]

Architect decomposition (LiteLLM on Kubernetes)

Coverage survey: rw-cli-codecollection has no existing LiteLLM-specific CodeBundles. Related patterns include HTTP API diagnostics (for example k8s-jaeger-http-query), Redis workload checks (k8s-redis-healthcheck—often used as LiteLLM cache/store), and generic Kubernetes health bundles.

Proposed CodeBundles (design specs filed as child issues):

1. k8s-litellm-proxy-health — API liveness/readiness, models/routes, optional deep /health probe, integration health, optional Service/Endpoints correlation. Uses PROXY_BASE_URL (ClusterIP/DNS or kubectl port-forward to the proxy Service).

2. k8s-litellm-spend-governance — Spend logs, global spend vs threshold, key/user/team budgets and rate limits, aggregated failure/blocked-request signals.

Child issues

• [design-spec] k8s-litellm-proxy-health #88 — [design-spec] k8s-litellm-proxy-health
• [design-spec] k8s-litellm-spend-governance #89 — [design-spec] k8s-litellm-spend-governance

Scorer: Scaffold paths issue-87/scaffold-bundles/k8s-litellm-proxy-health and k8s-litellm-spend-governance (copied from k8s-namespace-healthcheck for naming/structure validation) scored 93/100 (python -m scorer.score from codebundle-farm, threshold 70).

SRE implementation should replace scaffolds with real tasks/scripts per specs.

[rw-codebundle-agent]

✅ Architect agent completed. Check child issues for design specs.