quickcart/dynatrace-workflows/workshop-predictive-pvc-usage.workflow.json

{
  "title": "[WORKSHOP] Predictive PVC USage",
  "description": "",
  "schemaVersion": 4,
  "trigger": {},
  "result": null,
  "type": "STANDARD",
  "input": {},
  "hourlyExecutionLimit": 1000,
  "guide": null,
  "tasks": {
    "send_email_1": {
      "name": "send_email_1",
      "input": {
        "cc": [],
        "to": [
          "mark.bley@dynatrace.com"
        ],
        "bcc": [],
        "content": "PVC Is going to be full in the next hours !",
        "subject": "PVC Is going to be full in the next hours !"
      },
      "action": "dynatrace.email:send-email",
      "position": {
        "x": 1,
        "y": 3
      },
      "conditions": {
        "custom": "{{result(\"simple_prediction\")}}",
        "states": {
          "simple_prediction": "SUCCESS"
        }
      },
      "description": "Send email",
      "predecessors": [
        "simple_prediction"
      ]
    },
    "run_javascript_1": {
      "name": "run_javascript_1",
      "input": {
        "script": "import { execution } from '@dynatrace-sdk/automation-utils';\n\nconst TASK_ID = 'predict_disk_capacity';\n\n// 2 GiB in bytes\nconst THRESHOLD_BYTES = 2n * 1024n * 1024n * 1024n; // 2147483648\n\n// Choose which forecast series to evaluate: \"upper\" (recommended), \"point\", or \"lower\"\nconst SERIES_TO_CHECK = 'upper';\n\nfunction toGiB(bytesNumber) {\n  return bytesNumber / (1024 ** 3);\n}\n\nfunction parseIntervalNs(interval) {\n  // interval comes as a string like \"180000000000\" (ns)\n  try {\n    return BigInt(interval);\n  } catch {\n    return null;\n  }\n}\n\nfunction addMs(isoStart, ms) {\n  return new Date(new Date(isoStart).getTime() + ms).toISOString();\n}\n\nfunction scanSeries(values, thresholdBigInt) {\n  let max = -Infinity;\n  let firstIdx = -1;\n  let firstVal = null;\n\n  for (let i = 0; i < (values?.length ?? 0); i++) {\n    const v = values[i];\n    if (typeof v === 'number' && Number.isFinite(v)) {\n      if (v > max) max = v;\n      if (firstIdx < 0 && BigInt(Math.trunc(v)) >= thresholdBigInt) {\n        firstIdx = i;\n        firstVal = v;\n      }\n    }\n  }\n\n  return { max, firstIdx, firstVal };\n}\n\nfunction pickForecastRecord(item) {\n  // Your payload has: item.timeSeriesDataWithPredictions.records[0]\n  // Use first record; adjust if you ever have multiple records per item.\n  return item?.timeSeriesDataWithPredictions?.records?.[0] ?? null;\n}\n\nfunction extractIdentity(item, rec) {\n  // Prefer the prediction record fields; fallback to analyzedTimeSeriesQuery fields.\n  const analyzedRec = item?.analyzedTimeSeriesQuery?.expression?.records?.[0] ?? {};\n\n  return {\n    cluster:\n      rec?.['k8s.cluster.name'] ??\n      analyzedRec?.['k8s.cluster.name'] ??\n      'unknown-cluster',\n    pvc:\n      rec?.['k8s.persistent_volume_claim.name'] ??\n      analyzedRec?.['k8s.persistent_volume_claim.name'] ??\n      'unknown-pvc',\n  };\n}\n\nexport default async function ({ execution_id }) {\n  const exe = await execution(execution_id);\n\n  const predResult = await exe.result(TASK_ID);\n  // In your snippet you did: const result = predResult['result'];\n  // Here we handle both shapes gracefully:\n  const payload = predResult?.result ?? predResult;\n  const outputs = payload?.output ?? [];\n\n  const breaches = [];\n\n  for (const item of outputs) {\n    const rec = pickForecastRecord(item);\n    if (!rec) continue;\n\n    const { cluster, pvc } = extractIdentity(item, rec);\n\n    const series =\n      (SERIES_TO_CHECK === 'upper' && rec?.['dt.davis.forecast:upper']) ||\n      (SERIES_TO_CHECK === 'point' && rec?.['dt.davis.forecast:point']) ||\n      (SERIES_TO_CHECK === 'lower' && rec?.['dt.davis.forecast:lower']) ||\n      // fallback preference order\n      rec?.['dt.davis.forecast:upper'] ||\n      rec?.['dt.davis.forecast:point'] ||\n      rec?.['dt.davis.forecast:lower'] ||\n      [];\n\n    const { max, firstIdx, firstVal } = scanSeries(series, THRESHOLD_BYTES);\n\n    const start = rec?.timeframe?.start; // e.g. \"2026-03-04T00:33Z\"\n    const intervalNs = parseIntervalNs(rec?.interval);\n    const intervalMs = intervalNs ? Number(intervalNs / 1_000_000n) : null;\n\n    const firstCrossingTime =\n      firstIdx >= 0 && start && intervalMs != null\n        ? addMs(start, firstIdx * intervalMs)\n        : null;\n\n    if (firstIdx >= 0) {\n      breaches.push({\n        cluster,\n        pvc,\n        analysisStatus: item?.analysisStatus,\n        forecastQualityAssessment: item?.forecastQualityAssessment,\n        checkedSeries:\n          series === rec?.['dt.davis.forecast:upper']\n            ? 'dt.davis.forecast:upper'\n            : series === rec?.['dt.davis.forecast:point']\n              ? 'dt.davis.forecast:point'\n              : series === rec?.['dt.davis.forecast:lower']\n                ? 'dt.davis.forecast:lower'\n                : 'fallback',\n        thresholdBytes: Number(THRESHOLD_BYTES),\n        thresholdGiB: 2,\n        firstCrossing: {\n          index: firstIdx,\n          time: firstCrossingTime,\n          valueBytes: firstVal,\n          valueGiB: firstVal != null ? toGiB(firstVal) : null,\n        },\n        maxForecast: {\n          valueBytes: max,\n          valueGiB: Number.isFinite(max) ? toGiB(max) : null,\n        },\n      });\n    }\n  }\n\n  const summary = {\n    thresholdBytes: Number(THRESHOLD_BYTES),\n    thresholdGiB: 2,\n    checkedSeriesPreference: SERIES_TO_CHECK,\n    totalSeries: outputs.length,\n    breachCount: breaches.length,\n    okCount: outputs.length - breaches.length,\n    breaches,\n  };\n\n  console.log(JSON.stringify(summary, null, 2));\n\n  // OPTIONAL: Fail the workflow if any series breaches the threshold\n  // if (breaches.length > 0) {\n  //   throw new Error(`Forecast exceeds 2GiB for ${breaches.length} series`);\n  // }\n\n  return summary;\n}"
      },
      "action": "dynatrace.automations:run-javascript",
      "active": false,
      "position": {
        "x": -1,
        "y": 2
      },
      "conditions": {
        "states": {
          "predict_disk_capacity": "OK"
        }
      },
      "description": "Run custom JavaScript code.",
      "predecessors": [
        "predict_disk_capacity"
      ]
    },
    "simple_prediction": {
      "name": "simple_prediction",
      "input": {
        "script": "import { execution } from '@dynatrace-sdk/automation-utils';\n\nconst TASK_ID = 'predict_disk_capacity';\nconst THRESHOLD = 2 * 1024 * 1024 * 1024; // 2GiB in bytes\n\nexport default async function ({ execution_id }) {\n  const exe = await execution(execution_id);\n  const predResult = await exe.result(TASK_ID);\n  const result = predResult.result;\n\n  for (const item of result.output) {\n    const rec = item.timeSeriesDataWithPredictions.records[0];\n    const series = rec['dt.davis.forecast:upper'] || rec['dt.davis.forecast:point'];\n\n    for (const v of series) {\n      if (v != null && v >= THRESHOLD) return true;\n    }\n  }\n\n  return false;\n}"
      },
      "action": "dynatrace.automations:run-javascript",
      "position": {
        "x": 1,
        "y": 2
      },
      "conditions": {
        "states": {
          "predict_disk_capacity": "OK"
        }
      },
      "description": "Run custom JavaScript code.",
      "predecessors": [
        "predict_disk_capacity"
      ]
    },
    "predict_disk_capacity": {
      "name": "predict_disk_capacity",
      "input": {
        "body": {
          "nPaths": 200,
          "forecastOffset": 0,
          "timeSeriesData": "timeseries { avg(dt.kubernetes.persistentvolumeclaim.used), value.A = avg(dt.kubernetes.persistentvolumeclaim.used, scalar: true) }, by: { k8s.cluster.name, k8s.persistent_volume_claim.name }",
          "forecastHorizon": 200,
          "generalParameters": {
            "timeframe": {
              "endTime": "now",
              "startTime": "now()-6h"
            },
            "logVerbosity": "WARNING",
            "resolveDimensionalQueryData": false
          },
          "coverageProbability": 0.9,
          "applyZeroLowerBoundHeuristic": true
        },
        "analyzerName": "dt.statistics.GenericForecastAnalyzer"
      },
      "action": "dynatrace.davis.workflow.actions:davis-analyze",
      "position": {
        "x": 0,
        "y": 1
      },
      "description": "Execute a customizable AI/ML task using data analyzers",
      "predecessors": []
    }
  }
}