SERVER-67290 Support repair on clustered collections with missing _mdb_catalog

(cherry picked from commit bbfaa9d)

8 files changed, 51 insertions, 16 deletions

diff --git a/src/mongo/db/storage/SConscript b/src/mongo/db/storage/SConscript
index 45931873ba3..6081de4589c 100644
--- a/src/mongo/db/storage/SConscript
+++ b/src/mongo/db/storage/SConscript
@@ -623,6 +623,7 @@ env.Library(
     LIBDEPS_PRIVATE=[
         '$BUILD_DIR/mongo/db/audit',
         '$BUILD_DIR/mongo/db/catalog/catalog_helpers',
+        '$BUILD_DIR/mongo/db/catalog/clustered_collection_options',
         '$BUILD_DIR/mongo/db/catalog/index_catalog',
         '$BUILD_DIR/mongo/db/multitenancy',
         '$BUILD_DIR/mongo/db/resumable_index_builds_idl',
diff --git a/src/mongo/db/storage/durable_catalog.h b/src/mongo/db/storage/durable_catalog.h
index 49db4546918..d41e8173883 100644
--- a/src/mongo/db/storage/durable_catalog.h
+++ b/src/mongo/db/storage/durable_catalog.h
@@ -123,11 +123,12 @@ public:
     /**
      * Create an entry in the catalog for an orphaned collection found in the
      * storage engine. Return the generated ns of the collection.
-     * Note that this function does not recreate the _id index on the collection because it does not
-     * have access to index catalog.
+     * Note that this function does not recreate the _id index on the for non-clustered collections
+     * because it does not have access to index catalog.
      */
     virtual StatusWith<std::string> newOrphanedIdent(OperationContext* opCtx,
-                                                     std::string ident) = 0;
+                                                     std::string ident,
+                                                     const CollectionOptions& optionsWithUUID) = 0;
 
     virtual std::string getFilesystemPathForDb(const std::string& dbName) const = 0;
 
diff --git a/src/mongo/db/storage/durable_catalog_impl.cpp b/src/mongo/db/storage/durable_catalog_impl.cpp
index 13b86f9af07..c4ef8527e52 100644
--- a/src/mongo/db/storage/durable_catalog_impl.cpp
+++ b/src/mongo/db/storage/durable_catalog_impl.cpp
@@ -554,8 +554,8 @@ bool DurableCatalogImpl::isCollectionIdent(StringData ident) const {
         ident.find("collection/") != std::string::npos;
 }
 
-StatusWith<std::string> DurableCatalogImpl::newOrphanedIdent(OperationContext* opCtx,
-                                                             std::string ident) {
+StatusWith<std::string> DurableCatalogImpl::newOrphanedIdent(
+    OperationContext* opCtx, std::string ident, const CollectionOptions& optionsWithUUID) {
     // The collection will be named local.orphan.xxxxx.
     std::string identNs = ident;
     std::replace(identNs.begin(), identNs.end(), '-', '_');
@@ -563,9 +563,6 @@ StatusWith<std::string> DurableCatalogImpl::newOrphanedIdent(OperationContext* o
     NamespaceString nss(NamespaceString(NamespaceString::kOrphanCollectionDb,
                                         NamespaceString::kOrphanCollectionPrefix + identNs));
 
-    // Generate a new UUID for the orphaned collection.
-    CollectionOptions optionsWithUUID;
-    optionsWithUUID.uuid.emplace(UUID::gen());
     BSONObj obj;
     {
         BSONObjBuilder b;
diff --git a/src/mongo/db/storage/durable_catalog_impl.h b/src/mongo/db/storage/durable_catalog_impl.h
index 873733d58b4..bedaf4a79af 100644
--- a/src/mongo/db/storage/durable_catalog_impl.h
+++ b/src/mongo/db/storage/durable_catalog_impl.h
@@ -98,7 +98,9 @@ public:
         return _rs;
     }
 
-    StatusWith<std::string> newOrphanedIdent(OperationContext* opCtx, std::string ident);
+    StatusWith<std::string> newOrphanedIdent(OperationContext* opCtx,
+                                             std::string ident,
+                                             const CollectionOptions& optionsWithUUID);
 
     std::string getFilesystemPathForDb(const std::string& dbName) const;
 
diff --git a/src/mongo/db/storage/kv/kv_engine.h b/src/mongo/db/storage/kv/kv_engine.h
index 25f52f0e40f..304a695308e 100644
--- a/src/mongo/db/storage/kv/kv_engine.h
+++ b/src/mongo/db/storage/kv/kv_engine.h
@@ -422,6 +422,13 @@ public:
     }
 
     /**
+     * Returns the 'KeyFormat' tied to 'ident'.
+     */
+    virtual KeyFormat getKeyFormat(OperationContext* opCtx, StringData ident) const {
+        MONGO_UNREACHABLE;
+    }
+
+    /**
      * The destructor will never be called from mongod, but may be called from tests.
      * Engines may assume that this will only be called in the case of clean shutdown, even if
      * cleanShutdown() hasn't been called.
diff --git a/src/mongo/db/storage/storage_engine_impl.cpp b/src/mongo/db/storage/storage_engine_impl.cpp
index 255f80efec1..6e9c57de195 100644
--- a/src/mongo/db/storage/storage_engine_impl.cpp
+++ b/src/mongo/db/storage/storage_engine_impl.cpp
@@ -35,6 +35,7 @@
 
 #include "mongo/db/audit.h"
 #include "mongo/db/catalog/catalog_control.h"
+#include "mongo/db/catalog/clustered_collection_util.h"
 #include "mongo/db/catalog/collection_catalog.h"
 #include "mongo/db/catalog/collection_catalog_helper.h"
 #include "mongo/db/catalog_raii.h"
@@ -201,19 +202,36 @@ void StorageEngineImpl::loadCatalog(OperationContext* opCtx, LastShutdownState l
                     // If the catalog does not have information about this
                     // collection, we create an new entry for it.
                     WriteUnitOfWork wuow(opCtx);
-                    StatusWith<std::string> statusWithNs = _catalog->newOrphanedIdent(opCtx, ident);
+
+                    auto keyFormat = _engine->getKeyFormat(opCtx, ident);
+                    bool isClustered = keyFormat == KeyFormat::String;
+                    CollectionOptions optionsWithUUID;
+                    optionsWithUUID.uuid.emplace(UUID::gen());
+                    if (isClustered) {
+                        optionsWithUUID.clusteredIndex =
+                            clustered_util::makeDefaultClusteredIdIndex();
+                    }
+
+                    StatusWith<std::string> statusWithNs =
+                        _catalog->newOrphanedIdent(opCtx, ident, optionsWithUUID);
+
                     if (statusWithNs.isOK()) {
                         wuow.commit();
                         auto orphanCollNs = statusWithNs.getValue();
                         LOGV2(22247,
                               "Successfully created an entry in the catalog for orphaned "
                               "collection",
-                              "namespace"_attr = orphanCollNs);
-                        LOGV2_WARNING(22265,
-                                      "Collection does not have an _id index. Please manually "
-                                      "build the index",
-                                      "namespace"_attr = orphanCollNs);
-
+                              "namespace"_attr = orphanCollNs,
+                              "options"_attr = optionsWithUUID);
+
+                        if (!isClustered) {
+                            // The _id index is already implicitly created on collections clustered
+                            // by _id.
+                            LOGV2_WARNING(22265,
+                                          "Collection does not have an _id index. Please manually "
+                                          "build the index",
+                                          "namespace"_attr = orphanCollNs);
+                        }
                         StorageRepairObserver::get(getGlobalServiceContext())
                             ->benignModification(str::stream() << "Orphan collection created: "
                                                                << statusWithNs.getValue());
diff --git a/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.cpp b/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.cpp
index 575335efe75..048cac044d2 100644
--- a/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.cpp
+++ b/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.cpp
@@ -2756,4 +2756,11 @@ Status WiredTigerKVEngine::reconfigureLogging() {
     return wtRCToStatus(_conn->reconfigure(_conn, verboseConfig.c_str()), nullptr);
 }
 
+KeyFormat WiredTigerKVEngine::getKeyFormat(OperationContext* opCtx, StringData ident) const {
+
+    const std::string wtTableConfig =
+        uassertStatusOK(WiredTigerUtil::getMetadataCreate(opCtx, "table:{}"_format(ident)));
+    return wtTableConfig.find("key_format=u") != string::npos ? KeyFormat::String : KeyFormat::Long;
+}
+
 }  // namespace mongo
diff --git a/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.h b/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.h
index 688db855b74..add1d9a9b29 100644
--- a/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.h
+++ b/src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.h
@@ -383,6 +383,8 @@ public:
 
     Status reconfigureLogging() override;
 
+    KeyFormat getKeyFormat(OperationContext* opCtx, StringData ident) const override;
+
 private:
     class WiredTigerSessionSweeper;