OpenShift OVN-Kubernetes Benchmark MCP Server

# metrics.yml - OVN Kubernetes Container Metrics Configuration # This file defines the Prometheus queries for collecting OVN container metrics # If this file is not found or empty, hardcoded PromQL queries will be used metrics: # CPU Usage Metrics - query: 'rate(container_cpu_usage_seconds_total{pod=~"ovnkube-master.*|ovnkube-node.*", container!="POD", container!=""}[5m]) * 100' metricName: CONTAINER_CPU_USAGE_PERCENT unit: percentage description: "Container CPU usage percentage over 5 minutes" # Memory Usage Metrics - query: 'container_memory_rss{pod=~"ovnkube-master.*|ovnkube-node.*", container!="POD", container!=""}' metricName: CONTAINER_MEMORY_RSS_BYTES unit: bytes description: "Container memory RSS usage" - query: 'container_memory_working_set_bytes{pod=~"ovnkube-master.*|ovnkube-node.*", container!="POD", container!=""}' metricName: CONTAINER_MEMORY_WORKING_SET_BYTES unit: bytes description: "Container memory working set" # OVN Database Size Metrics (as specified in requirements) - query: 'ovn_db_db_size_bytes{db_name=~"OVN_Northbound"}' metricName: NBDB_DB_SIZE_BYTES unit: bytes description: "OVN Northbound database size" - query: 'ovn_db_db_size_bytes{db_name=~"OVN_Southbound"}' metricName: SBDB_DB_SIZE_BYTES unit: bytes description: "OVN Southbound database size" # OVN Container-Specific Memory Metrics (as specified in requirements) - query: 'sum by(pod) (container_memory_rss{pod=~"ovnkube-master.*|ovnkube-node.*", container=~"nbdb"})' metricName: OVN_DB_DB_MEMROY_SIZE_BYTES unit: bytes description: "NBDB container memory usage" - query: 'sum by(pod) (container_memory_rss{pod=~"ovnkube-master.*|ovnkube-node.*", container=~"sbdb"})' metricName: OVN_DB_DB_MEMROY_SIZE_BYTES unit: bytes description: "SBDB container memory usage" - query: 'sum by(pod) (container_memory_rss{pod=~"ovnkube-master.*|ovnkube-node.*", container=~"northd"})' metricName: OVN_DB_DB_MEMROY_SIZE_BYTES unit: bytes description: "Northd container memory usage" - query: 'sum by(pod) (container_memory_rss{pod=~"ovnkube-master.*|ovnkube-node.*", container=~"ovnkube-controller"})' metricName: OVN_KUBE_CONTROLLER_MEMROY_SIZE_BYTES unit: bytes description: "OVN Kube controller memory usage" - query: 'sum by(pod) (container_memory_rss{pod=~"ovnkube-master.*|ovnkube-node.*", container=~"ovn-controller"})' metricName: OVN_CONTROLLER_MEMROY_SIZE_BYTES unit: bytes description: "OVN controller memory usage" - query: topk(10, ovnkube_controller_ready_duration_seconds) metricName: ovnkube_controller_ready_duration_seconds unit: seconds description: "Time taken for ovnkube-controller to become ready" component: controller - query: topk(10, ovnkube_node_ready_duration_second s) metricName: ovnkube_node_ready_duration_seconds unit: seconds description: "Time taken for ovnkube-node to become ready" component: node - query: ovnkube_controller_sync_duration_seconds{namespace="openshift-ovn-kubernetes"} metricName: ovnkube_controller_sync_duration_seconds unit: seconds description: "Duration of controller sync operations" component: controller # Additional metrics can be added here - query: histogram_quantile(0.95, rate(ovnkube_controller_sync_duration_seconds_bucket[5m])) metricName: ovnkube_controller_sync_duration_p95 unit: seconds description: "95th percentile of controller sync duration over 5 minutes" component: controller - query: histogram_quantile(0.99, sum by (pod, le) (rate(ovnkube_controller_pod_creation_latency_seconds_bucket[$interval]))) > 0 metricName: Pod_Annotation_Latency_p99 - query: histogram_quantile(0.99, sum(rate(ovnkube_node_cni_request_duration_seconds_bucket{command="ADD"}[$interval])) by (pod,le)) > 0 metricName: CNI_Request_ADD_Latency_p99 - query: histogram_quantile(0.99, sum(rate(ovnkube_node_cni_request_duration_seconds_bucket{command="DEL"}[$interval])) by (pod,le)) > 0 metricName: CNI_Request_DEL_Latency_p99 - query: histogram_quantile(0.99, sum by(pod, le) (rate(ovnkube_controller_pod_first_seen_lsp_created_duration_seconds_bucket[2m]))) metricName: Pod_creation_latency_first_seen_lsp_p99 - query: histogram_quantile(0.99, sum by(pod, le) (rate(ovnkube_controller_pod_lsp_created_port_binding_duration_seconds_bucket[2m]))) metricName: pod_creation_latency_port_binding_p99 - query: rate(ovnkube_controller_sync_service_latency_seconds_sum[2m]) metricName: sync_service_latency - query: histogram_quantile(0.9, sum by(pod, le) (rate(ovnkube_controller_sync_service_latency_seconds_bucket[2m]))) metricName: sync_service_latency_p99 - query: histogram_quantile(0.99, sum by(pod, le) (rate(ovnkube_controller_network_programming_duration_seconds_bucket[2m]))) metricName: apply_network_config_pod_duration_p99 - query: histogram_quantile(0.99, sum by(service, le) (rate(ovnkube_controller_network_programming_duration_seconds_bucket[2m]))) metricName: apply_network_config_service_duration_p99 # Additional Custom Metrics (examples) # - query: 'ovn_nb_logical_switch_ports' # metricName: OVN_NB_LOGICAL_SWITCH_PORTS # unit: count # description: "Number of logical switch ports in NB database" # - query: 'ovn_sb_datapath_binding' # metricName: OVN_SB_DATAPATH_BINDING # unit: count # description: "Number of datapath bindings in SB database" # - query: 'rate(container_network_receive_bytes_total{pod=~"ovnkube.*"}[5m])' # metricName: OVN_CONTAINER_NETWORK_RX_BYTES_RATE # unit: bytes_per_second # description: "Container network receive rate" # Configuration notes: # - All queries should target OVN-related pods using pod=~"ovnkube-master.*|ovnkube-node.*" # - Use meaningful metricName values that will be used in the analysis # - Specify appropriate units: bytes, percentage, count, bytes_per_second, etc. # - Include helpful descriptions for documentation # - Comment out unused metrics or add custom ones as needed