Cilium

By enabling this toolset, HolmesGPT will be able to interact with Cilium CNI and Hubble network observability, providing comprehensive network troubleshooting and monitoring capabilities for Kubernetes clusters.

Prerequisites

1. Cilium CNI installed in your Kubernetes cluster
2. Hubble network observability enabled (for Hubble tools)
3. cilium CLI tool installed and configured
4. hubble CLI tool installed and configured (for Hubble tools)
5. Appropriate RBAC permissions for Cilium resources

Configuration

Holmes CLIRobusta Helm Chart

First, ensure your tools are properly configured:

# Verify Cilium is accessible
cilium version
cilium status

# Verify Hubble is accessible (if using Hubble tools)
hubble version
hubble status

Then add the following to ~/.holmes/config.yaml. Create the file if it doesn't exist:

toolsets:
  cilium/core:
    enabled: true
  hubble/observability:
    enabled: true

After making changes to your configuration, run:

holmes toolset refresh

holmes:
  toolsets:
    cilium/core:
      enabled: true
    hubble/observability:
      enabled: true

Update your Helm values and run a Helm upgrade:

helm upgrade robusta robusta/robusta --values=generated_values.yaml --set clusterName=<YOUR_CLUSTER_NAME>

Advanced Configuration

You can configure additional settings for enhanced functionality:

toolsets:
  cilium/core:
    enabled: true
    config:
      timeout: 60  # Command timeout in seconds
  hubble/observability:
    enabled: true
    config:
      max_flows: 10000  # Maximum flows to observe
      timeout: 120      # Extended timeout for flow monitoring

Capabilities

Cilium/Core Toolset

Tool Name	Description
cilium_status	Display overall Cilium agent status and health
cilium_status_verbose	Display detailed Cilium agent status with verbose output
cilium_version	Show Cilium version information
cilium_config	Display current Cilium configuration
cilium_endpoint_list	List all Cilium endpoints (pods with networking)
cilium_endpoint_get	Get detailed information about a specific endpoint by ID
cilium_endpoint_health	Check health status of all endpoints
cilium_endpoint_logs	Show logs for a specific endpoint
cilium_service_list	List all Cilium services and load balancer mappings
cilium_service_get	Get detailed information about a specific service by ID
cilium_loadbalancer_list	List load balancer backend mappings
cilium_policy_get	Get network policies applied to an endpoint
cilium_policy_trace	Trace policy decisions for traffic between source and destination
cilium_policy_trace_verbose	Detailed policy trace with verbose output
cilium_monitor	Monitor Cilium datapath events in real-time
cilium_monitor_verbose	Monitor datapath events with verbose output
cilium_node_list	List all nodes in the Cilium cluster
cilium_clustermesh_status	Display cluster mesh status for multi-cluster networking
cilium_bpf_map_list	List all BPF maps used by Cilium
cilium_bpf_endpoint_list	List endpoints from BPF maps
cilium_debuginfo	Generate debug information bundle for troubleshooting

Hubble/Observability Toolset

Tool Name	Description
hubble_observe	Observe network flows in real-time (last 1000 flows)
hubble_observe_follow	Follow network flows in real-time as they happen
hubble_observe_namespace	Observe flows for a specific namespace
hubble_observe_pod	Observe flows to/from a specific pod
hubble_observe_since	Observe flows since a specific time
hubble_observe_http	Observe HTTP traffic flows
hubble_observe_dns	Observe DNS queries and responses
hubble_observe_kafka	Observe Kafka protocol traffic
hubble_observe_grpc	Observe gRPC traffic flows
hubble_observe_drops	Show only dropped network flows (policy denials, etc.)
hubble_observe_denied	Show flows denied by network policies
hubble_observe_service	Observe flows to/from a specific service
hubble_observe_port	Observe flows on a specific port
hubble_observe_from_pod	Observe flows originating from a specific pod
hubble_observe_to_pod	Observe flows destined to a specific pod
hubble_observe_between_namespaces	Observe flows between two specific namespaces
hubble_observe_json	Output flow observations in JSON format for detailed analysis
hubble_status	Display Hubble server status and configuration
hubble_list_nodes	List nodes available for flow observation
hubble_observe_flows_summary	Get a summary of recent network flows with basic statistics
hubble_observe_security_events	Observe security-related network events and policy violations
hubble_observe_l7_denied	Show L7 (application-layer) traffic that was denied

Use Cases

Network Connectivity Troubleshooting

• Pod-to-Pod Communication: Use cilium_endpoint_list and hubble_observe_between_namespaces to diagnose connectivity issues
• Service Resolution: Check cilium_service_list and observe DNS flows with hubble_observe_dns
• Load Balancing Issues: Analyze cilium_loadbalancer_list and service traffic patterns

Network Policy Debugging

• Policy Violations: Use hubble_observe_denied to see blocked traffic
• Policy Tracing: Use cilium_policy_trace to understand policy decisions
• Security Events: Monitor with hubble_observe_security_events

Performance Analysis

• Network Monitoring: Real-time observation with cilium_monitor and hubble_observe_follow
• Flow Analysis: Detailed traffic patterns with hubble_observe_json
• Component Health: Status checking with cilium_status_verbose

Multi-Cluster Networking

• Cluster Mesh: Status and connectivity with cilium_clustermesh_status
• Cross-Cluster Flows: Monitor inter-cluster traffic patterns

Common Troubleshooting Scenarios

"My pods can't communicate"

1. Check Cilium agent health: cilium_status
2. Verify endpoints are healthy: cilium_endpoint_health
3. Observe blocked flows: hubble_observe_drops
4. Trace policy decisions: cilium_policy_trace

"DNS resolution is failing"

1. Monitor DNS queries: hubble_observe_dns
2. Check endpoint configuration: cilium_endpoint_get
3. Verify service mappings: cilium_service_list

"Network policies aren't working"

1. View denied flows: hubble_observe_denied
2. Trace policy evaluation: cilium_policy_trace_verbose
3. Check applied policies: cilium_policy_get

"Load balancing issues"

1. Examine service backends: cilium_service_get
2. Check BPF load balancer state: cilium_loadbalancer_list
3. Monitor service traffic: hubble_observe_service