diff --git a/Prometheus-Grafana/grafana/provisioning/dashboards/airflow-cluster-dashboard.json b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow-cluster-dashboard.json new file mode 100644 index 00000000..ab0fd4a5 --- /dev/null +++ b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow-cluster-dashboard.json @@ -0,0 +1,1611 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": { + "type": "datasource", + "uid": "grafana" + }, + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "fiscalYearStartMonth": 0, + "graphTooltip": 0, + "links": [], + "liveNow": false, + "panels": [ + { + "collapsed": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 43, + "panels": [], + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "refId": "A" + } + ], + "title": "Scheduler", + "type": "row" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "", + "fieldConfig": { + "defaults": { + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 8, + "x": 0, + "y": 1 + }, + "id": 17, + "links": [], + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_scheduler_heartbeat{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Scheduler heartbeat", + "type": "stat" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fieldConfig": { + "defaults": { + "links": [], + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 8, + "y": 1 + }, + "id": 4, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagbag_size{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Dagbag size", + "type": "stat" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken to scan and import all DAG files once", + "fieldConfig": { + "defaults": { + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 11, + "y": 1 + }, + "id": 14, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dag_processing_total_parse_time{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "DAG processing total parse time", + "type": "stat" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Number of errors from trying to parse DAG files", + "fieldConfig": { + "defaults": { + "links": [], + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 15, + "y": 1 + }, + "id": 6, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dag_processing_import_errors{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Dagbag import errors", + "type": "stat" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Zombie tasks killed", + "fieldConfig": { + "defaults": { + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 5, + "x": 19, + "y": 1 + }, + "id": 29, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_zombies_killed{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Zombies killed", + "type": "stat" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Overall task instances successes", + "fieldConfig": { + "defaults": { + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 0, + "y": 5 + }, + "id": 35, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_ti_successes{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Total successful tasks", + "type": "stat" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "", + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 8, + "x": 4, + "y": 5 + }, + "hiddenSeries": false, + "id": 39, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "rate(af_agg_operator_successes{airflow_id=\"$airflow_id\"}[1m]) * 60", + "interval": "", + "legendFormat": "{{operator_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Operator success rate per minuite", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "ops", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Total failures of given operator", + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 8, + "x": 12, + "y": 5 + }, + "hiddenSeries": false, + "id": 41, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "rate(af_agg_operator_failures{airflow_id=\"$airflow_id\"}[1m]) * 60", + "interval": "", + "legendFormat": "{{operator_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Operator failure rate per minute", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "ops", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "show": false + } + ], + "yaxis": { + "align": false + } + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fieldConfig": { + "defaults": { + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 5 + }, + "id": 37, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_ti_failures{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Total failed tasks", + "type": "stat" + }, + { + "collapsed": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 11 + }, + "id": 45, + "panels": [], + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "refId": "A" + } + ], + "title": "Executor Pool", + "type": "row" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 16, + "x": 0, + "y": 12 + }, + "hiddenSeries": false, + "id": 56, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "rate(af_agg_job_start{airflow_id=\"$airflow_id\"}[1m]) * 60", + "instant": false, + "interval": "", + "legendFormat": "{{job_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Jobs started per minute", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fieldConfig": { + "defaults": { + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 12 + }, + "id": 10, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_executor_open_slots{airflow_id=\"$airflow_id\"}", + "instant": false, + "interval": "1", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "title": "Executor open slots", + "type": "stat" + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fieldConfig": { + "defaults": { + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 12 + }, + "id": 53, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_executor_queued_tasks{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Executor queued tasks", + "type": "stat" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 16, + "x": 0, + "y": 18 + }, + "hiddenSeries": false, + "id": 21, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "rate(af_agg_job_end{airflow_id=\"$airflow_id\"}[1m]) * 60", + "interval": "", + "legendFormat": "{{job_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Jobs ended per minute", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "fieldConfig": { + "defaults": { + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 18 + }, + "id": 55, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "10.0.3", + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_executor_running_tasks{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Executor running tasks", + "type": "stat" + }, + { + "collapsed": true, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 24 + }, + "id": 47, + "panels": [], + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "refId": "A" + } + ], + "title": "DAG stats", + "type": "row" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken for a DagRun to reach success state", + "fieldConfig": { + "defaults": { + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 24, + "x": 0, + "y": 25 + }, + "hiddenSeries": false, + "id": 25, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_duration_success{airflow_id=\"$airflow_id\"} * 1000", + "format": "time_series", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Success DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "logBase": 1, + "show": true + }, + { + "format": "short", + "logBase": 1, + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken for a DagRun to reach failed state", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green" + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 33 + }, + "hiddenSeries": false, + "id": 49, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_duration_failed{airflow_id=\"$airflow_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Failed DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Sum of seconds taken to finish a tasks aggregated by DAG", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green" + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 40 + }, + "hiddenSeries": false, + "id": 51, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "sum by (dag_id, quantile) (af_agg_dag_task_duration{airflow_id=\"$airflow_id\"}) * 1000", + "format": "time_series", + "instant": false, + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken to check DAG dependencies", + "fieldConfig": { + "defaults": { + "custom": { + "filterable": false + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green" + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 46 + }, + "hiddenSeries": false, + "id": 19, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_dependency_check{airflow_id=\"$airflow_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "DAG run dependency check time", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds of delay between the scheduled DagRun start date and the actual DagRun start date", + "fieldConfig": { + "defaults": { + "custom": {}, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 53 + }, + "hiddenSeries": false, + "id": 27, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_schedule_delay{airflow_id=\"$airflow_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "DAG run schedule delay", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + } + ], + "refresh": "5s", + "schemaVersion": 38, + "style": "dark", + "tags": [], + "templating": { + "list": [ + { + "current": { + "selected": false, + "text": "airflow", + "value": "airflow" + }, + "definition": "label_values(airflow_id)", + "hide": 0, + "includeAll": false, + "label": "Airflow Instance Id", + "multi": false, + "name": "airflow_id", + "options": [], + "query": "label_values(airflow_id)", + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tagsQuery": "", + "type": "query", + "useTags": false + } + ] + }, + "time": { + "from": "now-1h", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ] + }, + "timezone": "", + "title": "Airflow cluster dashboard", + "uid": "lFXqBGxWk", + "version": 1, + "weekStart": "" + } diff --git a/Prometheus-Grafana/grafana/provisioning/dashboards/airflow-dag-dashboard.json b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow-dag-dashboard.json new file mode 100644 index 00000000..b7142cdd --- /dev/null +++ b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow-dag-dashboard.json @@ -0,0 +1,581 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": { + "type": "datasource", + "uid": "grafana" + }, + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "fiscalYearStartMonth": 0, + "graphTooltip": 2, + "links": [], + "liveNow": false, + "panels": [ + { + "collapsed": true, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 47, + "panels": [], + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "refId": "A" + } + ], + "title": "DAG stats", + "type": "row" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken for a DagRun to reach success state", + "fieldConfig": { + "defaults": { + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 24, + "x": 0, + "y": 1 + }, + "hiddenSeries": false, + "id": 25, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_duration_success{dag_id=\"$dag_id\"} * 1000", + "format": "time_series", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Success DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "logBase": 1, + "show": true + }, + { + "format": "short", + "logBase": 1, + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken for a DagRun to reach failed state", + "fieldConfig": { + "defaults": { + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 9 + }, + "hiddenSeries": false, + "id": 49, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_duration_failed{dag_id=\"$dag_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "Failed DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Sum of seconds taken to finish a tasks aggregated by DAG", + "fieldConfig": { + "defaults": { + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 16 + }, + "hiddenSeries": false, + "id": 51, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "sum by (dag_id, quantile) (af_agg_dag_task_duration{dag_id=\"$dag_id\"}) * 1000", + "format": "time_series", + "instant": false, + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds taken to check DAG dependencies", + "fieldConfig": { + "defaults": { + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 22 + }, + "hiddenSeries": false, + "id": 19, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "10.0.3", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_dependency_check{dag_id=\"$dag_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "DAG run dependency check time", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "description": "Seconds of delay between the scheduled DagRun start date and the actual DagRun start date", + "fieldConfig": { + "defaults": { + "custom": {}, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 29 + }, + "hiddenSeries": false, + "id": 27, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "datasource": { + "type": "prometheus", + "uid": "P1809F7CD0C75ACF3" + }, + "expr": "af_agg_dagrun_schedule_delay{dag_id=\"$dag_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeRegions": [], + "title": "DAG run schedule delay", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "mode": "time", + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + }, + { + "format": "short", + "logBase": 1, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false + } + } + ], + "refresh": "5s", + "schemaVersion": 38, + "style": "dark", + "tags": [], + "templating": { + "list": [ + { + "current": { + "selected": false, + "text": "latest_only", + "value": "latest_only" + }, + "definition": "label_values(dag_id)", + "hide": 0, + "includeAll": false, + "label": "DAG ID", + "multi": false, + "name": "dag_id", + "options": [], + "query": { + "query": "label_values(dag_id)", + "refId": "PrometheusVariableQueryEditor-VariableQuery" + }, + "refresh": 1, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tagsQuery": "", + "type": "query", + "useTags": false + } + ] + }, + "time": { + "from": "now-1h", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ] + }, + "timezone": "", + "title": "Airflow DAG dashboard", + "uid": "qBWJRGPMz", + "version": 1, + "weekStart": "" + } diff --git a/Prometheus-Grafana/grafana/provisioning/dashboards/airflow_grafana_dashboard.json b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow_grafana_dashboard.json new file mode 100644 index 00000000..41880205 --- /dev/null +++ b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow_grafana_dashboard.json @@ -0,0 +1,646 @@ +{ + "__inputs": [ + { + "name": "DS_PROMETHEUS", + "label": "Prometheus", + "description": "", + "type": "datasource", + "pluginId": "prometheus", + "pluginName": "Prometheus" + } + ], + "__requires": [ + { + "type": "grafana", + "id": "grafana", + "name": "Grafana", + "version": "7.1.5" + }, + { + "type": "panel", + "id": "graph", + "name": "Graph", + "version": "" + }, + { + "type": "datasource", + "id": "prometheus", + "name": "Prometheus", + "version": "1.0.0" + } + ], + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "gnetId": null, + "graphTooltip": 0, + "id": null, + "links": [], + "panels": [ + { + "aliasColors": {}, + "bars": true, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": { + "align": null + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + } + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 0 + }, + "hiddenSeries": false, + "id": 6, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": false, + "linewidth": 1, + "nullPointMode": "null", + "percentage": false, + "pluginVersion": "7.1.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "increase(airflow_ti_successes[1m])", + "instant": false, + "interval": "", + "legendFormat": "Success", + "refId": "A" + }, + { + "expr": "increase(airflow_ti_failures[1m])", + "interval": "", + "legendFormat": "Failure", + "refId": "B" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Task Instances Successes & Failures", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:753", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:754", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + } + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 12, + "x": 0, + "y": 7 + }, + "hiddenSeries": false, + "id": 4, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "percentage": false, + "pluginVersion": "7.1.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(airflow_scheduler_heartbeat[10m])*60", + "interval": "", + "legendFormat": "Heartbeats per Minute", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Scheduler Heartbeat", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:519", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:520", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 7 + }, + "hiddenSeries": false, + "id": 10, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "percentage": false, + "pluginVersion": "7.1.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "airflow_scheduler_tasks_executable", + "interval": "", + "legendFormat": "Executable", + "refId": "A" + }, + { + "expr": "airflow_scheduler_tasks_pending", + "interval": "", + "legendFormat": "Pending", + "refId": "B" + }, + { + "expr": "airflow_scheduler_tasks_running", + "interval": "", + "legendFormat": "Running", + "refId": "C" + }, + { + "expr": "airflow_scheduler_tasks_starving", + "interval": "", + "legendFormat": "Starving", + "refId": "D" + }, + { + "expr": "airflow_scheduler_tasks_killed_externally", + "interval": "", + "legendFormat": "Killed Externally", + "refId": "E" + }, + { + "expr": "airflow_scheduler_tasks_without_dagrun", + "interval": "", + "legendFormat": "Without DagRun", + "refId": "F" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Scheduler Tasks", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:629", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "$$hashKey": "object:630", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 9, + "w": 12, + "x": 0, + "y": 15 + }, + "hiddenSeries": false, + "id": 2, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "percentage": false, + "pluginVersion": "7.1.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "airflow_executor_open_slots", + "interval": "", + "legendFormat": "Open Slots", + "refId": "A" + }, + { + "expr": "airflow_executor_queued_tasks", + "interval": "", + "legendFormat": "Queued Tasks", + "refId": "B" + }, + { + "expr": "airflow_executor_running_tasks", + "interval": "", + "legendFormat": "Running Tasks", + "refId": "C" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Executor", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:720", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:721", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": "${DS_PROMETHEUS}", + "decimals": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + } + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 12, + "x": 12, + "y": 15 + }, + "hiddenSeries": false, + "id": 8, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": false, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "percentage": false, + "pluginVersion": "7.1.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "airflow_dagbag_size", + "interval": "", + "legendFormat": "DagBag Size", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DagBag Size", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "$$hashKey": "object:165", + "decimals": null, + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "$$hashKey": "object:166", + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "refresh": false, + "schemaVersion": 26, + "style": "dark", + "tags": [], + "templating": { + "list": [] + }, + "time": { + "from": "now-6h", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ] + }, + "timezone": "", + "title": "Airflow Dashboard", + "uid": "ZArfYGOGk", + "version": 39 +} diff --git a/Prometheus-Grafana/grafana/provisioning/dashboards/airflow_grafana_dashboard_V2.json b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow_grafana_dashboard_V2.json new file mode 100644 index 00000000..13a2d446 --- /dev/null +++ b/Prometheus-Grafana/grafana/provisioning/dashboards/airflow_grafana_dashboard_V2.json @@ -0,0 +1,130 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "gnetId": null, + "graphTooltip": 0, + "id": 9, + "links": [], + "panels": [ + { + "datasource": "Prometheus", + "description": "Airflow task duration. Filtered out example tasks that come with Airflow.", + "fieldConfig": { + "defaults": { + "color": { + "mode": "palette-classic" + }, + "custom": { + "axisLabel": "Duration", + "axisPlacement": "auto", + "barAlignment": 0, + "drawStyle": "line", + "fillOpacity": 10, + "gradientMode": "none", + "hideFrom": { + "graph": false, + "legend": false, + "tooltip": false + }, + "lineInterpolation": "linear", + "lineStyle": { + "fill": "solid" + }, + "lineWidth": 1, + "pointSize": 5, + "scaleDistribution": { + "type": "linear" + }, + "showPoints": "auto", + "spanNulls": true + }, + "mappings": [ + { + "from": "", + "id": 1, + "text": "", + "to": "", + "type": 1, + "value": "" + } + ], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "short" + }, + "overrides": [] + }, + "gridPos": { + "h": 19, + "w": 19, + "x": 0, + "y": 0 + }, + "id": 2, + "options": { + "graph": {}, + "legend": { + "calcs": [], + "displayMode": "list", + "placement": "bottom" + }, + "tooltipOptions": { + "mode": "single" + } + }, + "pluginVersion": "7.5.4", + "targets": [ + { + "exemplar": true, + "expr": "af_agg_dag_processing_last_run_seconds{dag_file =~ \"website_backups|website_security|git_backups\"}", + "format": "time_series", + "hide": false, + "instant": false, + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "title": "Airflow task durations", + "type": "timeseries" + } + ], + "schemaVersion": 27, + "style": "dark", + "tags": [], + "templating": { + "list": [] + }, + "time": { + "from": "now-30m", + "to": "now" + }, + "timepicker": {}, + "timezone": "", + "title": "Nunez family Airflow tasks duration", + "uid": "OaTTYQrMk", + "version": 1 +} diff --git a/Prometheus-Grafana/grafana/provisioning/dashboards/cluster-dashboard.json b/Prometheus-Grafana/grafana/provisioning/dashboards/cluster-dashboard.json new file mode 100644 index 00000000..dd2e00ba --- /dev/null +++ b/Prometheus-Grafana/grafana/provisioning/dashboards/cluster-dashboard.json @@ -0,0 +1,1620 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "gnetId": null, + "graphTooltip": 0, + "id": 2, + "iteration": 1612802652229, + "links": [], + "panels": [ + { + "collapsed": false, + "datasource": null, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 43, + "panels": [], + "title": "Scheduler", + "type": "row" + }, + { + "cacheTimeout": null, + "datasource": null, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 8, + "x": 0, + "y": 1 + }, + "id": 17, + "interval": null, + "links": [], + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_scheduler_heartbeat{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Scheduler heartbeat", + "type": "stat" + }, + { + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [], + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 3, + "x": 8, + "y": 1 + }, + "id": 4, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_dagbag_size{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Dagbag size", + "type": "stat" + }, + { + "datasource": null, + "description": "Seconds taken to scan and import all DAG files once", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 11, + "y": 1 + }, + "id": 14, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_dag_processing_total_parse_time{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "DAG processing total parse time", + "type": "stat" + }, + { + "datasource": null, + "description": "Number of errors from trying to parse DAG files", + "fieldConfig": { + "defaults": { + "custom": {}, + "links": [], + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 4, + "x": 15, + "y": 1 + }, + "id": 6, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "mean" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_dag_processing_import_errors{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Dagbag import errors", + "type": "stat" + }, + { + "datasource": null, + "description": "Zombie tasks killed", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 4, + "w": 5, + "x": 19, + "y": 1 + }, + "id": 29, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_zombies_killed{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Zombies killed", + "type": "stat" + }, + { + "datasource": null, + "description": "Overall task instances successes", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 0, + "y": 5 + }, + "id": 35, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_ti_successes{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Total successful tasks", + "type": "stat" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "", + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 8, + "x": 4, + "y": 5 + }, + "hiddenSeries": false, + "id": 39, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(af_agg_operator_successes{airflow_id=\"$airflow_id\"}[1m]) * 60", + "interval": "", + "legendFormat": "{{operator_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Operator success rate per minuite", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "ops", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Total failures of given operator", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + } + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 8, + "x": 12, + "y": 5 + }, + "hiddenSeries": false, + "id": 41, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(af_agg_operator_failures{airflow_id=\"$airflow_id\"}[1m]) * 60", + "interval": "", + "legendFormat": "{{operator_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Operator failure rate per minute", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": "ops", + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": false + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 5 + }, + "id": 37, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_ti_failures{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Total failed tasks", + "type": "stat" + }, + { + "collapsed": false, + "datasource": null, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 11 + }, + "id": 45, + "panels": [], + "title": "Executor Pool", + "type": "row" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 16, + "x": 0, + "y": 12 + }, + "hiddenSeries": false, + "id": 56, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(af_agg_job_start{airflow_id=\"$airflow_id\"}[1m]) * 60", + "instant": false, + "interval": "", + "legendFormat": "{{job_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Jobs started per minute", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 12 + }, + "id": 10, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_executor_open_slots{airflow_id=\"$airflow_id\"}", + "instant": false, + "interval": "1", + "intervalFactor": 1, + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Executor open slots", + "type": "stat" + }, + { + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 20, + "y": 12 + }, + "id": 53, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_executor_queued_tasks{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Executor queued tasks", + "type": "stat" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {} + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 16, + "x": 0, + "y": 18 + }, + "hiddenSeries": false, + "id": 21, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "rate(af_agg_job_end{airflow_id=\"$airflow_id\"}[1m]) * 60", + "interval": "", + "legendFormat": "{{job_name}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Jobs ended per minute", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "datasource": null, + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + } + }, + "overrides": [] + }, + "gridPos": { + "h": 6, + "w": 4, + "x": 16, + "y": 18 + }, + "id": 55, + "options": { + "colorMode": "value", + "graphMode": "area", + "justifyMode": "auto", + "orientation": "auto", + "reduceOptions": { + "calcs": [ + "lastNotNull" + ], + "fields": "", + "values": false + }, + "textMode": "auto" + }, + "pluginVersion": "7.3.5", + "targets": [ + { + "expr": "af_agg_executor_running_tasks{airflow_id=\"$airflow_id\"}", + "interval": "", + "legendFormat": "", + "refId": "A" + } + ], + "timeFrom": null, + "timeShift": null, + "title": "Executor running tasks", + "type": "stat" + }, + { + "collapsed": true, + "datasource": null, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 24 + }, + "id": 47, + "panels": [], + "title": "DAG stats", + "type": "row" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds taken for a DagRun to reach success state", + "fieldConfig": { + "defaults": { + "custom": { + "align": null, + "filterable": false + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + }, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 24, + "x": 0, + "y": 25 + }, + "hiddenSeries": false, + "id": 25, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_duration_success{airflow_id=\"$airflow_id\"} * 1000", + "format": "time_series", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Success DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds taken for a DagRun to reach failed state", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 33 + }, + "hiddenSeries": false, + "id": 49, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_duration_failed{airflow_id=\"$airflow_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Failed DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Sum of seconds taken to finish a tasks aggregated by DAG", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 40 + }, + "hiddenSeries": false, + "id": 51, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum by (dag_id, quantile) (af_agg_dag_task_duration{airflow_id=\"$airflow_id\"}) * 1000", + "format": "time_series", + "instant": false, + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds taken to check DAG dependencies", + "fieldConfig": { + "defaults": { + "custom": { + "align": null, + "filterable": false + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 46 + }, + "hiddenSeries": false, + "id": 19, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_dependency_check{airflow_id=\"$airflow_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DAG run dependency check time", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds of delay between the scheduled DagRun start date and the actual DagRun start date", + "fieldConfig": { + "defaults": { + "custom": {}, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 53 + }, + "hiddenSeries": false, + "id": 27, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_schedule_delay{airflow_id=\"$airflow_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DAG run schedule delay", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "refresh": "5s", + "schemaVersion": 26, + "style": "dark", + "tags": [], + "templating": { + "list": [ + { + "allValue": null, + "current": { + "selected": false, + "text": "airflow", + "value": "airflow" + }, + "datasource": "Prometheus", + "definition": "label_values(airflow_id)", + "error": null, + "hide": 0, + "includeAll": false, + "label": "Airflow Instance Id", + "multi": false, + "name": "airflow_id", + "options": [ + { + "selected": true, + "text": "airflow", + "value": "airflow" + } + ], + "query": "label_values(airflow_id)", + "refresh": 0, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + } + ] + }, + "time": { + "from": "now-1h", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ] + }, + "timezone": "", + "title": "Airflow cluster dashboard", + "uid": "lFXqBGxWk", + "version": 8 +} diff --git a/Prometheus-Grafana/grafana/provisioning/dashboards/dag-dashboard.json b/Prometheus-Grafana/grafana/provisioning/dashboards/dag-dashboard.json new file mode 100644 index 00000000..ab19d631 --- /dev/null +++ b/Prometheus-Grafana/grafana/provisioning/dashboards/dag-dashboard.json @@ -0,0 +1,670 @@ +{ + "annotations": { + "list": [ + { + "builtIn": 1, + "datasource": "-- Grafana --", + "enable": true, + "hide": true, + "iconColor": "rgba(0, 211, 255, 1)", + "name": "Annotations & Alerts", + "type": "dashboard" + } + ] + }, + "editable": true, + "gnetId": null, + "graphTooltip": 0, + "id": 23, + "iteration": 1612802909274, + "links": [], + "panels": [ + { + "collapsed": true, + "datasource": null, + "gridPos": { + "h": 1, + "w": 24, + "x": 0, + "y": 0 + }, + "id": 47, + "panels": [], + "title": "DAG stats", + "type": "row" + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds taken for a DagRun to reach success state", + "fieldConfig": { + "defaults": { + "custom": { + "align": null, + "filterable": false + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + }, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 8, + "w": 24, + "x": 0, + "y": 1 + }, + "hiddenSeries": false, + "id": 25, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_duration_success{dag_id=\"$dag_id\"} * 1000", + "format": "time_series", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Success DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": null, + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds taken for a DagRun to reach failed state", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "noValue": "0", + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 9 + }, + "hiddenSeries": false, + "id": 49, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_duration_failed{dag_id=\"$dag_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "Failed DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Sum of seconds taken to finish a tasks aggregated by DAG", + "fieldConfig": { + "defaults": { + "custom": {}, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + } + ] + }, + "unit": "s" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 16 + }, + "hiddenSeries": false, + "id": 51, + "legend": { + "alignAsTable": false, + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "sum by (dag_id, quantile) (af_agg_dag_task_duration{dag_id=\"$dag_id\"}) * 1000", + "format": "time_series", + "instant": false, + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DAG run duration", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "s", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds taken to check DAG dependencies", + "fieldConfig": { + "defaults": { + "custom": { + "align": null, + "filterable": false + }, + "mappings": [], + "thresholds": { + "mode": "absolute", + "steps": [ + { + "color": "green", + "value": null + }, + { + "color": "red", + "value": 80 + } + ] + }, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 7, + "w": 24, + "x": 0, + "y": 22 + }, + "hiddenSeries": false, + "id": 19, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_dependency_check{dag_id=\"$dag_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DAG run dependency check time", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "ms", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + }, + { + "aliasColors": {}, + "bars": false, + "dashLength": 10, + "dashes": false, + "datasource": null, + "description": "Seconds of delay between the scheduled DagRun start date and the actual DagRun start date", + "fieldConfig": { + "defaults": { + "custom": {}, + "unit": "ms" + }, + "overrides": [] + }, + "fill": 1, + "fillGradient": 0, + "gridPos": { + "h": 6, + "w": 24, + "x": 0, + "y": 29 + }, + "hiddenSeries": false, + "id": 27, + "legend": { + "avg": false, + "current": false, + "max": false, + "min": false, + "show": true, + "total": false, + "values": false + }, + "lines": true, + "linewidth": 1, + "nullPointMode": "null", + "options": { + "alertThreshold": true + }, + "percentage": false, + "pluginVersion": "7.3.5", + "pointradius": 2, + "points": false, + "renderer": "flot", + "seriesOverrides": [], + "spaceLength": 10, + "stack": false, + "steppedLine": false, + "targets": [ + { + "expr": "af_agg_dagrun_schedule_delay{dag_id=\"$dag_id\"} * 1000", + "interval": "", + "legendFormat": "{{dag_id}} p{{quantile}}", + "refId": "A" + } + ], + "thresholds": [], + "timeFrom": null, + "timeRegions": [], + "timeShift": null, + "title": "DAG run schedule delay", + "tooltip": { + "shared": true, + "sort": 0, + "value_type": "individual" + }, + "type": "graph", + "xaxis": { + "buckets": null, + "mode": "time", + "name": null, + "show": true, + "values": [] + }, + "yaxes": [ + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + }, + { + "format": "short", + "label": null, + "logBase": 1, + "max": null, + "min": "0", + "show": true + } + ], + "yaxis": { + "align": false, + "alignLevel": null + } + } + ], + "refresh": "5s", + "schemaVersion": 26, + "style": "dark", + "tags": [], + "templating": { + "list": [ + { + "allValue": null, + "current": { + "selected": true, + "text": "java_dag", + "value": "java_dag" + }, + "datasource": "Prometheus", + "definition": "label_values(dag_id)", + "error": null, + "hide": 0, + "includeAll": false, + "label": "DAG ID", + "multi": false, + "name": "dag_id", + "options": [ + { + "selected": false, + "text": "dbnd_spark_dag", + "value": "dbnd_spark_dag" + }, + { + "selected": false, + "text": "huge_dag", + "value": "huge_dag" + }, + { + "selected": true, + "text": "java_dag", + "value": "java_dag" + }, + { + "selected": false, + "text": "luigi_top_artists", + "value": "luigi_top_artists" + }, + { + "selected": false, + "text": "process_data_spark", + "value": "process_data_spark" + }, + { + "selected": false, + "text": "pyspark_dag", + "value": "pyspark_dag" + }, + { + "selected": false, + "text": "spark_dag", + "value": "spark_dag" + } + ], + "query": "label_values(dag_id)", + "refresh": 0, + "regex": "", + "skipUrlSync": false, + "sort": 0, + "tagValuesQuery": "", + "tags": [], + "tagsQuery": "", + "type": "query", + "useTags": false + } + ] + }, + "time": { + "from": "now-1h", + "to": "now" + }, + "timepicker": { + "refresh_intervals": [ + "5s", + "10s", + "30s", + "1m", + "5m", + "15m", + "30m", + "1h", + "2h", + "1d" + ] + }, + "timezone": "", + "title": "Airflow DAG dashboard", + "uid": "qBWJRGPMz", + "version": 4 +} diff --git a/Prometheus-Grafana/prometheus/prometheus.yml b/Prometheus-Grafana/prometheus/prometheus.yml index b400bfe2..0aed4142 100644 --- a/Prometheus-Grafana/prometheus/prometheus.yml +++ b/Prometheus-Grafana/prometheus/prometheus.yml @@ -80,3 +80,10 @@ scrape_configs: scrape_interval: 5s static_configs: - targets: ['10.12.1.14:8181'] + + - job_name: 'airflow' + # Override the global default and scrape targets from this job every 5 seconds. + scrape_interval: 5s + static_configs: + - targets: + - '10.12.1.14:9102'