From d92996e71f1dadb0cdf653531d11c602548dfc7a Mon Sep 17 00:00:00 2001
From: Gidi Meir Morris <github@gidi.io>
Date: Tue, 2 Feb 2021 11:28:29 +0000
Subject: [PATCH] support workload histograms that take into account overdue
 tasks

---
 .../monitoring/workload_statistics.test.ts    | 119 ++++++++++++++++++
 .../server/monitoring/workload_statistics.ts  |  13 +-
 2 files changed, 130 insertions(+), 2 deletions(-)

diff --git a/x-pack/plugins/task_manager/server/monitoring/workload_statistics.test.ts b/x-pack/plugins/task_manager/server/monitoring/workload_statistics.test.ts
index 21c9f429814cac..ed7c48d046dc2b 100644
--- a/x-pack/plugins/task_manager/server/monitoring/workload_statistics.test.ts
+++ b/x-pack/plugins/task_manager/server/monitoring/workload_statistics.test.ts
@@ -775,6 +775,125 @@ describe('padBuckets', () => {
       })
     ).toEqual([0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]);
   });
+
+  test('supports histogram buckets that begin in the past when tasks are overdue', async () => {
+    expect(
+      padBuckets(20, 3000, {
+        key: '2021-02-02T10:08:32.161Z-2021-02-02T10:09:32.161Z',
+        from: 1612260512161,
+        from_as_string: '2021-02-02T10:08:32.161Z',
+        to: 1612260572161,
+        to_as_string: '2021-02-02T10:09:32.161Z',
+        doc_count: 2,
+        histogram: {
+          buckets: [
+            {
+              key_as_string: '2021-02-02T10:08:30.000Z',
+              key: 1612260510000,
+              doc_count: 1,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [
+                  {
+                    key: '2s',
+                    doc_count: 1,
+                  },
+                ],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:33.000Z',
+              key: 1612260513000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:36.000Z',
+              key: 1612260516000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:39.000Z',
+              key: 1612260519000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:42.000Z',
+              key: 1612260522000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:45.000Z',
+              key: 1612260525000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:48.000Z',
+              key: 1612260528000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:51.000Z',
+              key: 1612260531000,
+              doc_count: 0,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [],
+              },
+            },
+            {
+              key_as_string: '2021-02-02T10:08:54.000Z',
+              key: 1612260534000,
+              doc_count: 1,
+              interval: {
+                doc_count_error_upper_bound: 0,
+                sum_other_doc_count: 0,
+                buckets: [
+                  {
+                    key: '60s',
+                    doc_count: 1,
+                  },
+                ],
+              },
+            },
+          ],
+        },
+      }).length
+      // we need to ensure overdue buckets don't cause us to over pad the timeline by adding additional
+      // buckets before and after the reported timeline
+    ).toEqual(20);
+  });
 });
 
 function setTaskTypeCount(
diff --git a/x-pack/plugins/task_manager/server/monitoring/workload_statistics.ts b/x-pack/plugins/task_manager/server/monitoring/workload_statistics.ts
index 8002ee44d01ff3..8bd22bd88cf410 100644
--- a/x-pack/plugins/task_manager/server/monitoring/workload_statistics.ts
+++ b/x-pack/plugins/task_manager/server/monitoring/workload_statistics.ts
@@ -244,10 +244,19 @@ export function padBuckets(
     const firstBucket = histogram.buckets[0].key;
     const lastBucket = histogram.buckets[histogram.buckets.length - 1].key;
 
-    const bucketsToPadBeforeFirstBucket = calculateBucketsBetween(firstBucket, from, pollInterval);
+    // detect when the first bucket is before the `from` so that we can take that into
+    // account by begining the timeline earlier
+    // This can happen when you have overdue tasks and Elasticsearch returns their bucket
+    // as begining before the `from`
+    const firstBucketStartsInThePast = firstBucket - from < 0;
+
+    const bucketsToPadBeforeFirstBucket = firstBucketStartsInThePast
+      ? []
+      : calculateBucketsBetween(firstBucket, from, pollInterval);
+
     const bucketsToPadAfterLast = calculateBucketsBetween(
       lastBucket + pollInterval,
-      to,
+      firstBucketStartsInThePast ? to - pollInterval : to,
       pollInterval
     );