SERVER-72595 Prevent query-based reopening from considering compressed buckets

3 files changed, 73 insertions, 43 deletions

diff --git a/jstests/core/timeseries/timeseries_reopened_bucket_insert.js b/jstests/core/timeseries/timeseries_reopened_bucket_insert.js
index 4a36277cc31..9bfcd73e8c5 100644
--- a/jstests/core/timeseries/timeseries_reopened_bucket_insert.js
+++ b/jstests/core/timeseries/timeseries_reopened_bucket_insert.js
@@ -215,40 +215,6 @@ const expectToReopenArchivedBuckets = function() {
     jsTestLog("Exiting expectToReopenArchivedBuckets.");
 }();
 
-const expectToReopenCompressedBuckets = function() {
-    jsTestLog("Entering expectToReopenCompressedBuckets...");
-    resetCollection();
-
-    let initialMeasurements = [];
-    for (let i = 0; i < 5; ++i) {
-        initialMeasurements.push({
-            [timeField]: ISODate("2022-08-26T19:19:00Z"),
-            [metaField]: "ReopenedBucket1",
-        });
-    }
-    const forward = {
-        [timeField]: ISODate("2022-08-27T19:19:00Z"),
-        [metaField]: "ReopenedBucket1",
-    };
-    const backward = {
-        [timeField]: ISODate("2022-08-26T19:19:00Z"),
-        [metaField]: "ReopenedBucket1",
-    };
-
-    for (let i = 0; i < initialMeasurements.length; ++i) {
-        checkIfBucketReopened(
-            initialMeasurements[i], /* willCreateBucket */ i == 0, /* willReopenBucket */ false);
-    }
-    // Time forwards will open a new bucket, and close and compress the old one.
-    checkIfBucketReopened(forward, /* willCreateBucket */ true, /* willReopenBucket */ false);
-    assert.eq(1, bucketsColl.find({"control.version": 2}).toArray().length);
-
-    // We expect to reopen the compressed bucket with time backwards.
-    checkIfBucketReopened(backward, /* willCreateBucket */ false, /* willReopenBucket */ true);
-
-    jsTestLog("Exiting expectToReopenCompressedBuckets.");
-}();
-
 const failToReopenNonSuitableBuckets = function() {
     jsTestLog("Entering failToReopenNonSuitableBuckets...");
     resetCollection();
@@ -294,7 +260,46 @@ const failToReopenNonSuitableBuckets = function() {
             "time": {"0": ISODate("2022-08-26T19:19:30Z")}
         }
     };
-
+    const compressedBucketDoc = {
+        "_id": ObjectId("05091c2c050b7495eaef4583"),
+        "control": {
+            "version": 2,
+            "min": {
+                "_id": ObjectId("63091c30138e9261fd70a903"),
+                "time": ISODate("2022-08-26T19:19:00Z")
+            },
+            "max": {
+                "_id": ObjectId("63091c30138e9261fd70a903"),
+                "time": ISODate("2022-08-26T19:19:30Z")
+            },
+            "closed": false
+        },
+        "meta": "NonSuitableBucket2",
+        "data": {
+            "_id": {"0": ObjectId("63091c30138e9261fd70a903")},
+            "time": {"0": ISODate("2022-08-26T19:19:30Z")}
+        }
+    };
+    const closedAndCompressedBucketDoc = {
+        "_id": ObjectId("06091c2c050b7495eaef4584"),
+        "control": {
+            "version": 2,
+            "min": {
+                "_id": ObjectId("63091c30138e9261fd70a903"),
+                "time": ISODate("2022-08-26T19:19:00Z")
+            },
+            "max": {
+                "_id": ObjectId("63091c30138e9261fd70a903"),
+                "time": ISODate("2022-08-26T19:19:30Z")
+            },
+            "closed": true
+        },
+        "meta": "NonSuitableBucket3",
+        "data": {
+            "_id": {"0": ObjectId("63091c30138e9261fd70a903")},
+            "time": {"0": ISODate("2022-08-26T19:19:30Z")}
+        }
+    };
     const year2000BucketDoc = {
         "_id": ObjectId("07091c2c050b7495eaef4585"),
         "control": {
@@ -340,6 +345,14 @@ const failToReopenNonSuitableBuckets = function() {
     // If an otherwise suitable bucket has the closed flag set, we expect to open a new bucket.
     checkIfBucketReopened(measurement1, /* willCreateBucket */ true, /* willReopenBucket */ false);
 
+    assert.commandWorked(bucketsColl.insert(compressedBucketDoc));
+    // If an otherwise suitable bucket is compressed, we expect to open a new bucket.
+    checkIfBucketReopened(measurement2, /* willCreateBucket */ true, /* willReopenBucket */ false);
+
+    assert.commandWorked(bucketsColl.insert(closedAndCompressedBucketDoc));
+    // If an otherwise suitable bucket is compressed and closed, we expect to open a new bucket.
+    checkIfBucketReopened(measurement3, /* willCreateBucket */ true, /* willReopenBucket */ false);
+
     assert.commandWorked(bucketsColl.insert(year2000BucketDoc));
     // If an otherwise suitable bucket has an incompatible time range with the measurement, we
     // expect to open a new bucket.
diff --git a/src/mongo/db/timeseries/bucket_catalog_helpers.cpp b/src/mongo/db/timeseries/bucket_catalog_helpers.cpp
index 6d3f0265a0f..13e174cd1bf 100644
--- a/src/mongo/db/timeseries/bucket_catalog_helpers.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog_helpers.cpp
@@ -135,6 +135,9 @@ BSONObj generateReopeningFilters(const Date_t& time,
                                  boost::optional<BSONElement> metadata,
                                  const std::string& controlMinTimePath,
                                  int64_t bucketMaxSpanSeconds) {
+    // The bucket must be uncompressed.
+    auto versionFilter = BSON(kControlVersionPath << kTimeseriesControlDefaultVersion);
+
     // The bucket cannot be closed (aka open for new measurements).
     auto closedFlagFilter =
         BSON("$or" << BSON_ARRAY(BSON(kControlClosedPath << BSON("$exists" << false))
@@ -158,7 +161,8 @@ BSONObj generateReopeningFilters(const Date_t& time,
     auto upperBound = BSON(controlMinTimePath << BSON("$gt" << measurementMaxDifference));
     auto timeRangeFilter = BSON("$and" << BSON_ARRAY(lowerBound << upperBound));
 
-    return BSON("$and" << BSON_ARRAY(closedFlagFilter << timeRangeFilter << metaFieldFilter));
+    return BSON("$and" << BSON_ARRAY(versionFilter << closedFlagFilter << timeRangeFilter
+                                                   << metaFieldFilter));
 }
 
 StatusWith<MinMax> generateMinMaxFromBucketDoc(const BSONObj& bucketDoc,
diff --git a/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp b/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp
index c056c07964a..0a5442d4080 100644
--- a/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp
@@ -379,7 +379,7 @@ TEST_F(BucketCatalogHelpersTest, FindSuitableBucketForMeasurements) {
                                        mongo::fromjson(
                                            R"({
             "_id":{"$oid":"629e1e680958e279dc29a517"},
-            "control":{"version":2,"min":{"_id":7,"time":{"$date":"2023-08-01T13:00:00Z"},"a":1},
+            "control":{"version":1,"min":{"_id":7,"time":{"$date":"2023-08-01T13:00:00Z"},"a":1},
                                    "max":{"_id":10,"time":{"$date":"2023-08-01T14:00:00Z"},"a":3},
                        "closed":false},
             "meta":3,
@@ -475,14 +475,25 @@ TEST_F(BucketCatalogHelpersTest, IncompatibleBucketsForNewMeasurements) {
     ASSERT(autoColl->getTimeseriesOptions() && autoColl->getTimeseriesOptions()->getMetaField());
     auto tsOptions = *autoColl->getTimeseriesOptions();
 
-    std::vector<BSONObj> bucketDocs = {// control.closed flag is true.
+    std::vector<BSONObj> bucketDocs = {// control.version indicates bucket is compressed.
+                                       mongo::fromjson(
+                                           R"({
+            "_id":{"$oid":"62e7e6ec27c28d338ab29200"},
+            "control":{"version":2,"min":{"_id":1,"time":{"$date":"2021-08-01T11:00:00Z"},"a":1},
+                                   "max":{"_id":3,"time":{"$date":"2021-08-01T12:00:00Z"},"a":3}},
+            "meta":1,
+            "data":{"time":{"0":{"$date":"2021-08-01T11:00:00Z"},
+                            "1":{"$date":"2021-08-01T11:00:00Z"},
+                            "2":{"$date":"2021-08-01T11:00:00Z"}},
+                    "a":{"0":1,"1":2,"2":3}}})"),
+                                       // control.closed flag is true.
                                        mongo::fromjson(
                                            R"(
             {"_id":{"$oid":"62e7eee4f33f295800073138"},
             "control":{"version":1,"min":{"_id":7,"time":{"$date":"2022-08-01T12:00:00Z"},"a":1},
                                    "max":{"_id":10,"time":{"$date":"2022-08-01T13:00:00Z"},"a":3},
                        "closed":true},
-            "meta":1,
+            "meta":2,
             "data":{"time":{"0":{"$date":"2022-08-01T12:00:00Z"},
                             "1":{"$date":"2022-08-01T12:00:00Z"},
                             "2":{"$date":"2022-08-01T12:00:00Z"}},
@@ -494,7 +505,7 @@ TEST_F(BucketCatalogHelpersTest, IncompatibleBucketsForNewMeasurements) {
             "control":{"version":2,"min":{"_id":7,"time":{"$date":"2023-08-01T13:00:00Z"},"a":1},
                                    "max":{"_id":10,"time":{"$date":"2023-08-01T14:00:00Z"},"a":3},
                        "closed":true},
-            "meta":2,
+            "meta":3,
             "data":{"time":{"0":{"$date":"2023-08-01T13:00:00Z"},
                             "1":{"$date":"2023-08-01T13:00:00Z"},
                             "2":{"$date":"2023-08-01T13:00:00Z"}},
@@ -505,11 +516,13 @@ TEST_F(BucketCatalogHelpersTest, IncompatibleBucketsForNewMeasurements) {
         _insertIntoBucketColl(doc);
     }
 
-    auto time1 = dateFromISOString("2022-08-01T12:30:00Z");
-    auto time2 = dateFromISOString("2023-08-01T13:30:00Z");
+    auto time1 = dateFromISOString("2021-08-01T11:30:00Z");
+    auto time2 = dateFromISOString("2022-08-01T12:30:00Z");
+    auto time3 = dateFromISOString("2023-08-01T13:30:00Z");
     std::vector<BSONObj> validMeasurementDocs = {
         BSON("_id" << 1 << _timeField << time1.getValue() << _metaField << 1),
-        BSON("_id" << 2 << _timeField << time2.getValue() << _metaField << 2)};
+        BSON("_id" << 2 << _timeField << time2.getValue() << _metaField << 2),
+        BSON("_id" << 3 << _timeField << time3.getValue() << _metaField << 3)};
 
     // Verify that even with matching meta fields and buckets with acceptable time ranges, if the
     // bucket is compressed and/or closed, we should not see it as a candid bucket for future