global: scrape_interval: 60s scrape_configs: - job_name: vmagent relabel_configs: - target_label: vm_account_id replacement: "100" static_configs: - targets: - localhost:8429 - bearer_token_file: /var/run/secrets/kubernetes.io/serviceaccount/token job_name: kubernetes-apiservers kubernetes_sd_configs: - role: endpointslice relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: keep regex: default;kubernetes;https source_labels: - __meta_kubernetes_namespace - __meta_kubernetes_service_name - __meta_kubernetes_endpoint_port_name scheme: https tls_config: ca_file: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt insecure_skip_verify: true - bearer_token_file: /var/run/secrets/kubernetes.io/serviceaccount/token job_name: kubernetes-nodes kubernetes_sd_configs: - role: node relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - replacement: kubernetes.default.svc:443 target_label: __address__ - regex: (.+) replacement: /api/v1/nodes/$1/proxy/metrics source_labels: - __meta_kubernetes_node_name target_label: __metrics_path__ scheme: https tls_config: ca_file: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt insecure_skip_verify: true - bearer_token_file: /var/run/secrets/kubernetes.io/serviceaccount/token job_name: kubernetes-nodes-cadvisor kubernetes_sd_configs: - role: node relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - replacement: kubernetes.default.svc:443 target_label: __address__ - regex: (.+) replacement: /api/v1/nodes/$1/proxy/metrics/cadvisor source_labels: - __meta_kubernetes_node_name target_label: __metrics_path__ scheme: https tls_config: ca_file: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt insecure_skip_verify: true - job_name: kubernetes-service-endpoints kubernetes_sd_configs: - role: endpointslice relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: drop regex: true source_labels: - __meta_kubernetes_pod_container_init - action: keep_if_equal source_labels: - __meta_kubernetes_service_annotation_prometheus_io_port - __meta_kubernetes_pod_container_port_number - action: keep regex: true source_labels: - __meta_kubernetes_service_annotation_prometheus_io_scrape - action: replace regex: (https?) source_labels: - __meta_kubernetes_service_annotation_prometheus_io_scheme target_label: __scheme__ - action: replace regex: (.+) source_labels: - __meta_kubernetes_service_annotation_prometheus_io_path target_label: __metrics_path__ - action: replace regex: ([^:]+)(?::\d+)?;(\d+) replacement: $1:$2 source_labels: - __address__ - __meta_kubernetes_service_annotation_prometheus_io_port target_label: __address__ - action: labelmap regex: __meta_kubernetes_service_label_(.+) - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_service_name target_label: kubernetes_name - action: replace source_labels: - __meta_kubernetes_pod_node_name target_label: kubernetes_node - job_name: kubernetes-service-endpoints-slow kubernetes_sd_configs: - role: endpointslice relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: drop regex: true source_labels: - __meta_kubernetes_pod_container_init - action: keep_if_equal source_labels: - __meta_kubernetes_service_annotation_prometheus_io_port - __meta_kubernetes_pod_container_port_number - action: keep regex: true source_labels: - __meta_kubernetes_service_annotation_prometheus_io_scrape_slow - action: replace regex: (https?) source_labels: - __meta_kubernetes_service_annotation_prometheus_io_scheme target_label: __scheme__ - action: replace regex: (.+) source_labels: - __meta_kubernetes_service_annotation_prometheus_io_path target_label: __metrics_path__ - action: replace regex: ([^:]+)(?::\d+)?;(\d+) replacement: $1:$2 source_labels: - __address__ - __meta_kubernetes_service_annotation_prometheus_io_port target_label: __address__ - action: labelmap regex: __meta_kubernetes_service_label_(.+) - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_service_name target_label: kubernetes_name - action: replace source_labels: - __meta_kubernetes_pod_node_name target_label: kubernetes_node scrape_interval: 60s scrape_timeout: 30s - job_name: kubernetes-services kubernetes_sd_configs: - role: service metrics_path: /probe params: module: - http_2xx relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: keep regex: true source_labels: - __meta_kubernetes_service_annotation_prometheus_io_probe - source_labels: - __address__ target_label: __param_target - replacement: blackbox target_label: __address__ - source_labels: - __param_target target_label: instance - action: labelmap regex: __meta_kubernetes_service_label_(.+) - source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - source_labels: - __meta_kubernetes_service_name target_label: kubernetes_name - job_name: kubernetes-pods kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: drop regex: true source_labels: - __meta_kubernetes_pod_container_init - action: keep_if_equal source_labels: - __meta_kubernetes_pod_annotation_prometheus_io_port - __meta_kubernetes_pod_container_port_number - action: keep regex: true source_labels: - __meta_kubernetes_pod_annotation_prometheus_io_scrape - action: replace regex: (.+) source_labels: - __meta_kubernetes_pod_annotation_prometheus_io_path target_label: __metrics_path__ - action: replace regex: ([^:]+)(?::\d+)?;(\d+) replacement: $1:$2 source_labels: - __address__ - __meta_kubernetes_pod_annotation_prometheus_io_port target_label: __address__ - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name - job_name: kubernetes-pods-jmx kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: drop regex: true source_labels: - __meta_kubernetes_pod_container_init - action: keep_if_equal source_labels: - __meta_kubernetes_pod_annotation_jmx_io_port - __meta_kubernetes_pod_container_port_number - action: keep regex: true source_labels: - __meta_kubernetes_pod_annotation_jmx_io_scrape - action: replace regex: (.+) source_labels: - __meta_kubernetes_pod_annotation_jmx_io_path target_label: __metrics_path__ - action: replace regex: ([^:]+)(?::\d+)?;(\d+) replacement: $1:$2 source_labels: - __address__ - __meta_kubernetes_pod_annotation_jmx_io_port target_label: __address__ - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name metric_relabel_configs: - if: '{__name__="kafka_producer_producer_topic_metrics_record_send_rate",topic="order_confirmed_ds_v2"}' action: drop - if: '{__name__="kafka_consumer_consumer_fetch_manager_metrics_records_consumed_rate",topic="order_confirmed_v3"}' action: drop - job_name: kube-state-metrics-k8s-dsgpu-prd-ase1 metrics_path: /metrics honor_timestamps: true scrape_interval: 60s scrape_timeout: 30s scheme: http static_configs: - targets: ['kube-state-metrics-dsgpu-prd.kube-system.svc.cluster.local:8080'] relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - job_name: airlfow-statsd-exporter metrics_path: /metrics honor_timestamps: true scrape_interval: 60s scrape_timeout: 30s scheme: http static_configs: - targets: ['prometheus-statsd-exporter.monitoring.svc.cluster.local:9102'] relabel_configs: - target_label: vm_account_id replacement: "100" - job_name: kubernetes-pods-telegraf kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: keep regex: '(.*9273)' source_labels: - __address__ - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name - job_name: kubernetes-pods-go-jmx metrics_path: /metrics kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: keep regex: '(.*8880)' source_labels: - __address__ - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name metric_relabel_configs: - if: '{__name__="kafka_producer_producer_topic_metrics_record_send_rate",topic="order_confirmed_ds_v2"}' action: drop - if: '{__name__="kafka_consumer_consumer_fetch_manager_metrics_records_consumed_rate",topic="order_confirmed_v3"}' action: drop - job_name: kubernetes-thirdparty-metrics metrics_path: /metrics kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: keep regex: '(.*8002)' source_labels: - __address__ - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name - job_name: kubernetes-pods-otel kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: drop regex: true source_labels: - __meta_kubernetes_pod_container_init - action: keep regex: true source_labels: - __meta_kubernetes_pod_annotation_otel_io_scrape - action: replace regex: (.+) source_labels: - __meta_kubernetes_pod_annotation_otel_io_path target_label: __metrics_path__ - action: replace regex: ([^:]+)(?::\d+)?;(\d+) replacement: $1:$2 source_labels: - __address__ - __meta_kubernetes_pod_annotation_otel_io_port target_label: __address__ - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name - job_name: 'telegraf-exporter' scrape_interval: 60s scrape_timeout: 5s gce_sd_configs: - project: 'meesho-datascience-prd-0622' port: 9273 zone: ["asia-southeast1-a", "asia-southeast1-b", "asia-southeast1-c"] relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_monitoring] regex: 'no' action: drop - source_labels: [__meta_gce_label_env] regex: '(DR|perf)' action: drop - source_labels: [__meta_gce_label_pod] regex: '(devops)' action: drop - source_labels: [__meta_gce_label_service] regex: '(kafka|hbase|supply|supply-internal|mongo)' action: drop - source_labels: [__meta_gce_label_telegraf] regex: '(enabled)' action: keep - source_labels: [__meta_gce_instance_name] target_label: name - source_labels: [__meta_gce_label_service] target_label: app - job_name: Flink01 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19900 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink02 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19901 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink03 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19902 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink04 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19903 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink05 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19904 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink06 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19905 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink07 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19906 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink08 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19907 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink09 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19908 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink10 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19909 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink11 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19910 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink12 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19911 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink13 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19912 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink14 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19913 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink15 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19914 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink16 metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19915 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(mlp-realtime-features-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink01Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19900 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink02Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19901 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink03Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19902 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink04Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19903 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink05Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19904 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink06Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19905 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink07Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19906 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink08Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19907 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink09Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19908 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink10Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19909 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink11Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19910 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink12Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19911 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink13Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19912 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink14Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19913 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink15Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19914 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: Flink16Reels metrics_path: / honor_timestamps: true stream_parse: true scrape_interval: 30s scrape_timeout: 30s scheme: http gce_sd_configs: - project: 'meesho-datascience-prd-0622' zone: 'asia-southeast1-a' port: 19915 relabel_configs: - target_label: vm_account_id replacement: "100" - source_labels: [__meta_gce_instance_status] regex: 'RUNNING' action: keep - source_labels: [__meta_gce_label_service] regex: '(realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_name] regex: '(reels-realtime-pipeline)' action: keep - source_labels: [__meta_gce_label_application] regex: '(hadoop)' action: keep - job_name: fluentd kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "100" - action: keep source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_name] regex: fluentd - action: replace source_labels: [__meta_kubernetes_pod_ip] target_label: __address__ regex: (.+) replacement: $1:24231 - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: replace source_labels: [__meta_kubernetes_pod_label_env] target_label: env - action: replace source_labels: [__meta_kubernetes_pod_label_bu] target_label: bu - action: replace source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance] target_label: fluentd_instance - job_name: kubernetes-pods-jmx-secured kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "200" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_node_label_(.+) - action: drop regex: true source_labels: - __meta_kubernetes_pod_container_init - action: keep_if_equal source_labels: - __meta_kubernetes_pod_annotation_jmx_io_port - __meta_kubernetes_pod_container_port_number - action: keep regex: true source_labels: - __meta_kubernetes_pod_annotation_jmx_io_scrape - action: replace regex: (.+) source_labels: - __meta_kubernetes_pod_annotation_jmx_io_path target_label: __metrics_path__ - action: replace regex: ([^:]+)(?::\d+)?;(\d+) replacement: $1:$2 source_labels: - __address__ - __meta_kubernetes_pod_annotation_jmx_io_port target_label: __address__ - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name metric_relabel_configs: - source_labels: [__name__,topic] regex: "(kafka_producer_producer_topic_metrics_record_send_rate;order_confirmed_ds_v2)|(kafka_consumer_consumer_fetch_manager_metrics_records_consumed_rate;order_confirmed_v3)" action: keep - job_name: kubernetes-pods-go-jmx-secured metrics_path: /metrics kubernetes_sd_configs: - role: pod relabel_configs: - target_label: vm_account_id replacement: "200" - target_label: cluster replacement: k8s-dsgpu-prd-ase1 - action: labelmap regex: __meta_kubernetes_pod_label_(.+) - action: keep regex: '(.*8880)' source_labels: - __address__ - action: replace source_labels: - __meta_kubernetes_namespace target_label: kubernetes_namespace - action: replace source_labels: - __meta_kubernetes_pod_name target_label: kubernetes_pod_name metric_relabel_configs: - source_labels: [__name__,topic] regex: "(kafka_producer_producer_topic_metrics_record_send_rate;order_confirmed_ds_v2)|(kafka_consumer_consumer_fetch_manager_metrics_records_consumed_rate;order_confirmed_v3)" action: keep