SERVER-54069 Added $graphLookup test suite for timeseriesSERVER-54990-refactor-TODO

3 files changed, 173 insertions, 15 deletions

diff --git a/etc/evergreen.yml b/etc/evergreen.yml
index 24966083b47..a63b854eb9c 100644
--- a/etc/evergreen.yml
+++ b/etc/evergreen.yml
@@ -5761,7 +5761,7 @@ tasks:
   - func: "do setup"
   - func: "run tests"
     vars:
-      resmoke_args: --suites=core --shellReadMode=legacy --shellWriteMode=compatibility --storageEngine=wiredTiger --excludeWithAnyTags=requires_find_command
+      resmoke_args: --suites=core --shellReadMode=legacy --shellWriteMode=compatibility --storageEngine=wiredTiger --excludeWithAnyTags=requires_find_command,requires_timeseries
 
 - <<: *task_template
   name: jsCore_auth
diff --git a/jstests/core/timeseries/libs/timeseries.js b/jstests/core/timeseries/libs/timeseries.js
index f1f4d3eb917..df52c18749a 100644
--- a/jstests/core/timeseries/libs/timeseries.js
+++ b/jstests/core/timeseries/libs/timeseries.js
@@ -29,6 +29,10 @@ var TimeseriesTest = class {
         return arr[Random.randInt(arr.length)];
     }
 
+    static getRandomUsage() {
+        return Random.randInt(101);
+    }
+
     /**
      * Generates time-series data based on the TSBS document-per-event format.
      *
@@ -38,10 +42,6 @@ var TimeseriesTest = class {
     static generateHosts(numHosts) {
         const hosts = new Array(numHosts);
 
-        const getRandomUsage = function() {
-            return Random.randInt(101);
-        };
-
         const regions = [
             "ap-northeast-1",
             "ap-southeast-1",
@@ -102,21 +102,22 @@ var TimeseriesTest = class {
             const regionIndex = Random.randInt(regions.length);
             hosts[i] = {
                 fields: {
-                    usage_guest: getRandomUsage(),
-                    usage_guest_nice: getRandomUsage(),
-                    usage_idle: getRandomUsage(),
-                    usage_iowait: getRandomUsage(),
-                    usage_irq: getRandomUsage(),
-                    usage_nice: getRandomUsage(),
-                    usage_softirq: getRandomUsage(),
-                    usage_steal: getRandomUsage(),
-                    usage_system: getRandomUsage(),
-                    usage_user: getRandomUsage(),
+                    usage_guest: TimeseriesTest.getRandomUsage(),
+                    usage_guest_nice: TimeseriesTest.getRandomUsage(),
+                    usage_idle: TimeseriesTest.getRandomUsage(),
+                    usage_iowait: TimeseriesTest.getRandomUsage(),
+                    usage_irq: TimeseriesTest.getRandomUsage(),
+                    usage_nice: TimeseriesTest.getRandomUsage(),
+                    usage_softirq: TimeseriesTest.getRandomUsage(),
+                    usage_steal: TimeseriesTest.getRandomUsage(),
+                    usage_system: TimeseriesTest.getRandomUsage(),
+                    usage_user: TimeseriesTest.getRandomUsage(),
                 },
                 tags: {
                     arch: TimeseriesTest.getRandomElem(["x64", "x86"]),
                     datacenter: TimeseriesTest.getRandomElem(dataCenters[regionIndex]),
                     hostname: "host_" + i,
+                    hostid: i,
                     os: TimeseriesTest.getRandomElem(
                         ["Ubuntu15.10", "Ubuntu16.10", "Ubuntu16.04LTS"]),
                     rack: Random.randInt(100).toString(),
diff --git a/jstests/core/timeseries/timeseries_graph_lookup.js b/jstests/core/timeseries/timeseries_graph_lookup.js
new file mode 100644
index 00000000000..29786acb8fb
--- /dev/null
+++ b/jstests/core/timeseries/timeseries_graph_lookup.js
@@ -0,0 +1,157 @@
+/**
+ * Verifies that time-series bucket collections work as expected with $graphLookup.
+ *
+ *
+ * @tags: [
+ *     assumes_unsharded_collection,
+ *     assumes_against_mongod_not_mongos,
+ *     requires_timeseries,
+ *     requires_fcv_49,
+ *     sbe_incompatible,
+ * ]
+ */
+(function() {
+"use strict";
+
+load("jstests/core/timeseries/libs/timeseries.js");
+
+if (!TimeseriesTest.timeseriesCollectionsEnabled(db.getMongo())) {
+    jsTestLog("Skipping test because the time-series collection feature flag is disabled");
+    return;
+}
+
+const testDB = db.getSiblingDB(jsTestName());
+assert.commandWorked(testDB.dropDatabase());
+const timeFieldName = "time";
+const hostIdFieldName = "hostid";
+const nonTimeseriesCollOption = null;
+const timeseriesCollOption = {
+    timeseries: {timeField: timeFieldName, metaField: hostIdFieldName}
+};
+const numHosts = 10;
+const numDocs = 200;
+
+Random.setRandomSeed();
+const hosts = TimeseriesTest.generateHosts(numHosts);
+
+let testFunc = function(collAOption, collBOption) {
+    // Prepares two collections. Each collection can be either a time-series or a non time-series
+    // collection, depending on collAOption/collBOption.
+    const collA = testDB.getCollection("a");
+    const collB = testDB.getCollection("b");
+    collA.drop();
+    collB.drop();
+    assert.commandWorked(testDB.createCollection(collA.getName(), collAOption));
+    assert.commandWorked(testDB.createCollection(collB.getName(), collBOption));
+    let entryCountPerHost = new Array(numHosts).fill(0);
+    let entryCountOver80AndExistsIdlePerHost = new Array(numHosts).fill(0);
+
+    // Inserts into collA, one entry per host.
+    for (let i = 0; i < numHosts; i++) {
+        let host = hosts[i];
+        assert.commandWorked(collA.insert({time: ISODate(), hostid: host.tags.hostid}));
+    }
+
+    // Inserts some random documents to collB. The 'idle' measurement is inserted only when usage is
+    // odd.
+    for (let i = 0; i < numDocs; i++) {
+        let host = TimeseriesTest.getRandomElem(hosts);
+        let usage = TimeseriesTest.getRandomUsage();
+        if (usage % 2) {
+            assert.commandWorked(collB.insert(
+                {time: ISODate(), hostid: host.tags.hostid, cpu: usage, idle: 100 - usage}));
+        } else {
+            assert.commandWorked(
+                collB.insert({time: ISODate(), hostid: host.tags.hostid, cpu: usage}));
+        }
+
+        // These counts are to test metaField match.
+        entryCountPerHost[host.tags.hostid]++;
+
+        // These counts are to test measurement fields match which are specified by $graphLookup's
+        // restrictSearchWithMatch.
+        if (usage > 80 && usage % 2) {
+            entryCountOver80AndExistsIdlePerHost[host.tags.hostid]++;
+        }
+    }
+
+    // Verifies that a meta field "hostid" works with $graphLookup.
+    let results = collA.aggregate([
+            {
+                $graphLookup: {
+                    from: collB.getName(),
+                    startWith: "$hostid",
+                    connectFromField: "hostid",
+                    connectToField: "hostid",
+                    as: "matchedB",
+                    maxDepth: 0
+                }
+            }, {
+                $project: {
+                    _id: 0,
+                    hostid: 1,
+                    matchedB: {
+                        $size: "$matchedB"
+                    }
+                }
+            },
+            {$sort: {hostid: 1}}
+        ]).toArray();
+
+    assert.eq(numHosts, results.length, results);
+
+    for (let i = 0; i < numHosts; i++) {
+        assert.eq({hostid: i, matchedB: entryCountPerHost[i]}, results[i], results);
+    }
+
+    // Verifies that measurement fields "cpu" and "idle" work with $graphLookup as expected.
+    results = collA.aggregate([
+        {
+            $graphLookup: {
+                from: collB.getName(),
+                startWith: "$hostid",
+                connectFromField: "hostid",
+                connectToField: "hostid",
+                as: "matchedB",
+                maxDepth: 0,
+                restrictSearchWithMatch: {
+                    cpu: {$gt: 80},             // Tests measurement "cpu".
+                    idle: {$exists: true}       // Tests the existence of measurement "idle".
+                }
+            }
+        }, {
+            $project: {
+                _id: 0,
+                hostid: 1,
+                matchedB: {
+                    $size: "$matchedB"
+                }
+            }
+        },
+        {$sort: {hostid: 1}}
+    ]).toArray();
+
+    assert.eq(numHosts, results.length, results);
+
+    for (let i = 0; i < numHosts; i++) {
+        let expectedCount = entryCountOver80AndExistsIdlePerHost[i];
+        assert.eq(
+            {hostid: i, matchedB: expectedCount}, results[i], entryCountOver80AndExistsIdlePerHost);
+    }
+};
+
+// Tests case #1: collA: non time-series, collB: time-series
+var collAOption = nonTimeseriesCollOption;
+var collBOption = timeseriesCollOption;
+testFunc(collAOption, collBOption);
+
+// Tests case #2: collA: time-series, collB: non time-series
+collAOption = timeseriesCollOption;
+collBOption = nonTimeseriesCollOption;
+testFunc(collAOption, collBOption);
+
+// Tests case #3: collA: time-series, collB: time-series
+collAOption = timeseriesCollOption;
+collBOption = timeseriesCollOption;
+testFunc(collAOption, collBOption);
+})();