SERVER-73978 Optimize shards retrieval in CheckMetadataConstistency command

1 files changed, 24 insertions, 16 deletions

diff --git a/src/mongo/s/commands/cluster_check_metadata_consistency_cmd.cpp b/src/mongo/s/commands/cluster_check_metadata_consistency_cmd.cpp
index 6babb16601c..4ecf7d49f78 100644
--- a/src/mongo/s/commands/cluster_check_metadata_consistency_cmd.cpp
+++ b/src/mongo/s/commands/cluster_check_metadata_consistency_cmd.cpp
@@ -42,7 +42,6 @@
 #include "mongo/s/request_types/sharded_ddl_commands_gen.h"
 #include "mongo/s/sharding_feature_flags_gen.h"
 
-
 namespace mongo {
 namespace {
 
@@ -52,6 +51,27 @@ namespace {
 // more than 8 decimal digits since the response is at most 16MB, and 16 * 1024 * 1024 < 1 * 10^8.
 static const int kPerDocumentOverheadBytesUpperBound = 10;
 
+/*
+ * Return the set of shards that are primaries for at least one database
+ */
+stdx::unordered_set<ShardId> getAllDbPrimaryShards(OperationContext* opCtx) {
+    static const std::vector<BSONObj> rawPipeline{fromjson(R"({
+        $group: {
+            _id: '$primary'
+        }
+    })")};
+    AggregateCommandRequest aggRequest{NamespaceString::kConfigDatabasesNamespace, rawPipeline};
+    auto aggResponse = Grid::get(opCtx)->catalogClient()->runCatalogAggregation(
+        opCtx, aggRequest, {repl::ReadConcernLevel::kMajorityReadConcern});
+
+    stdx::unordered_set<ShardId> shardIds;
+    shardIds.reserve(aggResponse.size());
+    for (auto&& responseEntry : aggResponse) {
+        shardIds.insert(responseEntry.firstElement().str());
+    }
+    return shardIds;
+}
+
 MetadataConsistencyCommandLevelEnum getCommandLevel(const NamespaceString& nss) {
     if (nss.isAdminDB()) {
         return MetadataConsistencyCommandLevelEnum::kClusterLevel;
@@ -102,25 +122,13 @@ public:
                 "cluster level mode must be run against the 'admin' database with {aggregate: 1}",
                 nss.isCollectionlessCursorNamespace());
 
-            const auto catalogClient = Grid::get(opCtx)->catalogClient();
-            // TODO: SERVER-73978: Retrieve directly from configsvr a list of shards with its
-            // corresponding primary databases.
-            const auto databases =
-                catalogClient->getAllDBs(opCtx, repl::ReadConcernLevel::kMajorityReadConcern);
-
             std::vector<std::pair<ShardId, BSONObj>> requests;
             ShardsvrCheckMetadataConsistency shardsvrRequest{nss};
-            shardsvrRequest.setDbName(nss.db());
             shardsvrRequest.setCursor(request().getCursor());
 
-            // Send a unique request per shard that is a primary shard for, at least, one database.
-            std::set<ShardId> shardIds;
-            for (const auto& db : databases) {
-                const auto insertionRes = shardIds.insert(db.getPrimary());
-                if (insertionRes.second) {
-                    // The shard was not in the set, so we need to send a request to it.
-                    requests.emplace_back(db.getPrimary(), shardsvrRequest.toBSON({}));
-                }
+            // Send a request to all shards that are primaries for at least one database
+            for (auto&& shardId : getAllDbPrimaryShards(opCtx)) {
+                requests.emplace_back(std::move(shardId), shardsvrRequest.toBSON({}));
             }
 
             // Send a request to the configsvr to check cluster metadata consistency.