SERVER-53567 Create conflict stash collections with same collation as temporary resharding collection

7 files changed, 104 insertions, 19 deletions

diff --git a/src/mongo/db/s/resharding/resharding_data_copy_util.cpp b/src/mongo/db/s/resharding/resharding_data_copy_util.cpp
index ad85a69d241..f870643e257 100644
--- a/src/mongo/db/s/resharding/resharding_data_copy_util.cpp
+++ b/src/mongo/db/s/resharding/resharding_data_copy_util.cpp
@@ -39,7 +39,9 @@
 
 namespace mongo::resharding::data_copy {
 
-void ensureCollectionExists(OperationContext* opCtx, const NamespaceString& nss) {
+void ensureCollectionExists(OperationContext* opCtx,
+                            const NamespaceString& nss,
+                            const CollectionOptions& options) {
     invariant(!opCtx->lockState()->isLocked());
     invariant(!opCtx->lockState()->inAWriteUnitOfWork());
 
@@ -50,7 +52,7 @@ void ensureCollectionExists(OperationContext* opCtx, const NamespaceString& nss)
         }
 
         WriteUnitOfWork wuow(opCtx);
-        coll.ensureDbExists()->createCollection(opCtx, nss);
+        coll.ensureDbExists()->createCollection(opCtx, nss, options);
         wuow.commit();
     });
 }
diff --git a/src/mongo/db/s/resharding/resharding_data_copy_util.h b/src/mongo/db/s/resharding/resharding_data_copy_util.h
index 22b96a1b26a..3374b8c7228 100644
--- a/src/mongo/db/s/resharding/resharding_data_copy_util.h
+++ b/src/mongo/db/s/resharding/resharding_data_copy_util.h
@@ -40,7 +40,14 @@ class OperationContext;
 
 namespace resharding::data_copy {
 
-void ensureCollectionExists(OperationContext* opCtx, const NamespaceString& nss);
+/**
+ * Creates the specified collection with the given options if the collection does not already exist.
+ * If the collection already exists, we do not compare the options because the resharding process
+ * will always use the same options for the same namespace.
+ */
+void ensureCollectionExists(OperationContext* opCtx,
+                            const NamespaceString& nss,
+                            const CollectionOptions& options);
 
 /**
  * Drops the specified collection or returns without error if the collection has already been
diff --git a/src/mongo/db/s/resharding/resharding_oplog_applier.cpp b/src/mongo/db/s/resharding/resharding_oplog_applier.cpp
index 263b8e05ec4..e2b4cc76667 100644
--- a/src/mongo/db/s/resharding/resharding_oplog_applier.cpp
+++ b/src/mongo/db/s/resharding/resharding_oplog_applier.cpp
@@ -496,14 +496,16 @@ Timestamp ReshardingOplogApplier::_clearAppliedOpsAndStoreProgress(OperationCont
     return lastAppliedTs;
 }
 
-NamespaceString ReshardingOplogApplier::ensureStashCollectionExists(OperationContext* opCtx,
-                                                                    const UUID& existingUUID,
-                                                                    const ShardId& donorShardId) {
+NamespaceString ReshardingOplogApplier::ensureStashCollectionExists(
+    OperationContext* opCtx,
+    const UUID& existingUUID,
+    const ShardId& donorShardId,
+    const CollectionOptions& options) {
     auto nss = NamespaceString{NamespaceString::kConfigDb,
                                "localReshardingConflictStash.{}.{}"_format(
                                    existingUUID.toString(), donorShardId.toString())};
 
-    resharding::data_copy::ensureCollectionExists(opCtx, nss);
+    resharding::data_copy::ensureCollectionExists(opCtx, nss, options);
     return nss;
 }
 
diff --git a/src/mongo/db/s/resharding/resharding_oplog_applier.h b/src/mongo/db/s/resharding/resharding_oplog_applier.h
index 71462dbc9a0..e0b7b4df5da 100644
--- a/src/mongo/db/s/resharding/resharding_oplog_applier.h
+++ b/src/mongo/db/s/resharding/resharding_oplog_applier.h
@@ -86,7 +86,8 @@ public:
 
     static NamespaceString ensureStashCollectionExists(OperationContext* opCtx,
                                                        const UUID& existingUUID,
-                                                       const ShardId& donorShardId);
+                                                       const ShardId& donorShardId,
+                                                       const CollectionOptions& options);
 
 private:
     using OplogBatch = std::vector<repl::OplogEntry>;
diff --git a/src/mongo/db/s/resharding/resharding_recipient_service.cpp b/src/mongo/db/s/resharding/resharding_recipient_service.cpp
index 548072bc3d4..cafff767d33 100644
--- a/src/mongo/db/s/resharding/resharding_recipient_service.cpp
+++ b/src/mongo/db/s/resharding/resharding_recipient_service.cpp
@@ -138,12 +138,35 @@ void createTemporaryReshardingCollectionLocally(OperationContext* opCtx,
     // Set the temporary resharding collection's UUID to the resharding UUID. Note that
     // BSONObj::addFields() replaces any fields that already exist.
     collOptions = collOptions.addFields(BSON("uuid" << reshardingUUID));
-
     CollectionOptionsAndIndexes optionsAndIndexes = {reshardingUUID, indexes, idIndex, collOptions};
     MigrationDestinationManager::cloneCollectionIndexesAndOptions(
         opCtx, reshardingNss, optionsAndIndexes);
 }
 
+std::vector<NamespaceString> ensureStashCollectionsExist(
+    OperationContext* opCtx,
+    const ChunkManager& cm,
+    const UUID& existingUUID,
+    std::vector<DonorShardMirroringEntry> donorShards) {
+    // Use the same collation for the stash collections as the temporary resharding collection
+    auto collator = cm.getDefaultCollator();
+    BSONObj collationSpec = collator ? collator->getSpec().toBSON() : BSONObj();
+
+    std::vector<NamespaceString> stashCollections;
+    stashCollections.reserve(donorShards.size());
+
+    {
+        CollectionOptions options;
+        options.collation = std::move(collationSpec);
+        for (const auto& donor : donorShards) {
+            stashCollections.emplace_back(ReshardingOplogApplier::ensureStashCollectionExists(
+                opCtx, existingUUID, donor.getId(), options));
+        }
+    }
+
+    return stashCollections;
+}
+
 }  // namespace resharding
 
 std::shared_ptr<repl::PrimaryOnlyService::Instance> ReshardingRecipientService::constructInstance(
@@ -420,16 +443,13 @@ ExecutorFuture<void> ReshardingRecipientService::RecipientStateMachine::
         return catalogCache->getShardedCollectionRoutingInfo(opCtx.get(), _recipientDoc.getNss());
     }();
 
-    std::vector<NamespaceString> stashCollections;
-    stashCollections.reserve(numDonors);
-
-    {
+    auto stashCollections = [&] {
         auto opCtx = cc().makeOperationContext();
-        for (const auto& donor : _recipientDoc.getDonorShardsMirroring()) {
-            stashCollections.emplace_back(ReshardingOplogApplier::ensureStashCollectionExists(
-                opCtx.get(), _recipientDoc.getExistingUUID(), donor.getId()));
-        }
-    }
+        return resharding::ensureStashCollectionsExist(opCtx.get(),
+                                                       sourceChunkMgr,
+                                                       _recipientDoc.getExistingUUID(),
+                                                       _recipientDoc.getDonorShardsMirroring());
+    }();
 
     size_t i = 0;
     auto futuresToWaitOn = std::move(_oplogFetcherFutures);
diff --git a/src/mongo/db/s/resharding/resharding_recipient_service.h b/src/mongo/db/s/resharding/resharding_recipient_service.h
index 045a55bc1a8..71939cdb06c 100644
--- a/src/mongo/db/s/resharding/resharding_recipient_service.h
+++ b/src/mongo/db/s/resharding/resharding_recipient_service.h
@@ -54,6 +54,12 @@ void createTemporaryReshardingCollectionLocally(OperationContext* opCtx,
                                                 const UUID& existingUUID,
                                                 Timestamp fetchTimestamp);
 
+std::vector<NamespaceString> ensureStashCollectionsExist(
+    OperationContext* opCtx,
+    const ChunkManager& cm,
+    const UUID& existingUUID,
+    std::vector<DonorShardMirroringEntry> donorShards);
+
 }  // namespace resharding
 
 class ReshardingRecipientService final : public repl::PrimaryOnlyService {
diff --git a/src/mongo/db/s/resharding/resharding_recipient_service_test.cpp b/src/mongo/db/s/resharding/resharding_recipient_service_test.cpp
index f8380457895..8a03ed223c6 100644
--- a/src/mongo/db/s/resharding/resharding_recipient_service_test.cpp
+++ b/src/mongo/db/s/resharding/resharding_recipient_service_test.cpp
@@ -34,6 +34,8 @@
 #include "mongo/bson/unordered_fields_bsonobj_comparator.h"
 #include "mongo/db/catalog_raii.h"
 #include "mongo/db/dbdirectclient.h"
+#include "mongo/db/query/collation/collator_factory_interface.h"
+#include "mongo/db/query/collation/collator_interface_mock.h"
 #include "mongo/db/repl/oplog.h"
 #include "mongo/db/repl/replication_coordinator_mock.h"
 #include "mongo/db/repl/storage_interface_impl.h"
@@ -133,13 +135,15 @@ public:
                                                         const NamespaceString& origNss,
                                                         const ShardKeyPattern& skey,
                                                         UUID uuid,
-                                                        OID epoch) {
+                                                        OID epoch,
+                                                        const BSONObj& collation = {}) {
         auto future = scheduleRoutingInfoForcedRefresh(tempNss);
 
         expectFindSendBSONObjVector(kConfigHostAndPort, [&]() {
             CollectionType coll(tempNss, epoch, Date_t::now(), uuid);
             coll.setKeyPattern(skey.getKeyPattern());
             coll.setUnique(false);
+            coll.setDefaultCollation(collation);
 
             TypeCollectionReshardingFields reshardingFields;
             reshardingFields.setUuid(uuid);
@@ -539,5 +543,48 @@ TEST_F(ReshardingRecipientServiceTest,
     verifyCollectionAndIndexes(kReshardingNss, kReshardingUUID, indexes);
 }
 
+TEST_F(ReshardingRecipientServiceTest, StashCollectionsHaveSameCollationAsReshardingCollection) {
+    auto shards = setupNShards(2);
+
+    std::unique_ptr<CollatorInterfaceMock> collator =
+        std::make_unique<CollatorInterfaceMock>(CollatorInterfaceMock::MockType::kReverseString);
+    auto collationSpec = collator->getSpec().toBSON();
+    auto srcChunkMgr = makeChunkManager(kOrigNss,
+                                        ShardKeyPattern(BSON("_id" << 1)),
+                                        std::move(collator),
+                                        false /* unique */,
+                                        {} /* splitPoints */);
+
+    // Create stash collections for both donor shards
+    auto stashCollections = resharding::ensureStashCollectionsExist(
+        operationContext(),
+        srcChunkMgr,
+        kOrigUUID,
+        {DonorShardMirroringEntry(ShardId("shard0"), true),
+         DonorShardMirroringEntry(ShardId("shard1"), true)});
+
+    // Verify that each stash collation has the collation we passed in above
+    {
+        auto opCtx = operationContext();
+
+        DBDirectClient client(opCtx);
+        auto collInfos = client.getCollectionInfos("config");
+        StringMap<BSONObj> nsToOptions;
+        for (const auto& coll : collInfos) {
+            nsToOptions[coll["name"].str()] = coll["options"].Obj();
+        }
+
+        for (const auto& coll : stashCollections) {
+            auto it = nsToOptions.find(coll.coll());
+            ASSERT(it != nsToOptions.end());
+            auto options = it->second;
+
+            ASSERT(options.hasField("collation"));
+            auto collation = options["collation"].Obj();
+            ASSERT_BSONOBJ_EQ(collationSpec, collation);
+        }
+    }
+}
+
 }  // namespace
 }  // namespace mongo