SERVER-71609 `splitOrMarkJumbo` must not over-split chunks

5 files changed, 28 insertions, 13 deletions

diff --git a/src/mongo/db/s/auto_split_vector_command.cpp b/src/mongo/db/s/auto_split_vector_command.cpp
index ba48fdffed8..2fa8ad9c5c3 100644
--- a/src/mongo/db/s/auto_split_vector_command.cpp
+++ b/src/mongo/db/s/auto_split_vector_command.cpp
@@ -89,7 +89,8 @@ public:
                                                                req.getKeyPattern(),
                                                                req.getMin(),
                                                                req.getMax(),
-                                                               req.getMaxChunkSizeBytes());
+                                                               req.getMaxChunkSizeBytes(),
+                                                               req.getLimit());
             return Response(std::move(splitPoints), continuation);
         }
 
diff --git a/src/mongo/db/s/config/sharding_catalog_manager_chunk_operations.cpp b/src/mongo/db/s/config/sharding_catalog_manager_chunk_operations.cpp
index d622162ef09..a9e62c5a472 100644
--- a/src/mongo/db/s/config/sharding_catalog_manager_chunk_operations.cpp
+++ b/src/mongo/db/s/config/sharding_catalog_manager_chunk_operations.cpp
@@ -1713,13 +1713,18 @@ void ShardingCatalogManager::splitOrMarkJumbo(OperationContext* opCtx,
                 Grid::get(opCtx)->getBalancerConfiguration()->getMaxChunkSizeBytes());
         }();
 
-        const auto splitPoints = uassertStatusOK(
+        // Limit the search to one split point: this code path is reached when a migration fails due
+        // to ErrorCodes::ChunkTooBig. In case there is a too frequent shard key, only select the
+        // next key in order to split the range in jumbo chunk + remaining range.
+        const int limit = 1;
+        auto splitPoints = uassertStatusOK(
             shardutil::selectChunkSplitPoints(opCtx,
                                               chunk.getShardId(),
                                               nss,
                                               cm.getShardKeyPattern(),
                                               ChunkRange(chunk.getMin(), chunk.getMax()),
-                                              maxChunkSizeBytes));
+                                              maxChunkSizeBytes,
+                                              limit));
 
         if (splitPoints.empty()) {
             LOGV2(21873,
@@ -1771,6 +1776,9 @@ void ShardingCatalogManager::splitOrMarkJumbo(OperationContext* opCtx,
             return;
         }
 
+        // Resize the vector because in multiversion scenarios the `autoSplitVector` command may end
+        // up ignoring the `limit` parameter and returning the whole list of split points.
+        splitPoints.resize(limit);
         uassertStatusOK(
             shardutil::splitChunkAtMultiplePoints(opCtx,
                                                   chunk.getShardId(),
diff --git a/src/mongo/s/request_types/auto_split_vector.idl b/src/mongo/s/request_types/auto_split_vector.idl
index be25b1c00ce..50e3c55ad92 100644
--- a/src/mongo/s/request_types/auto_split_vector.idl
+++ b/src/mongo/s/request_types/auto_split_vector.idl
@@ -75,3 +75,7 @@ commands:
         maxChunkSizeBytes:
             type: safeInt64
             description: "Max chunk size of the collection expressed in bytes"
+        limit:
+            type: int
+            description: "Max number of split points to look for"
+            optional: true
diff --git a/src/mongo/s/shard_util.cpp b/src/mongo/s/shard_util.cpp
index d0f64821611..09f8408de42 100644
--- a/src/mongo/s/shard_util.cpp
+++ b/src/mongo/s/shard_util.cpp
@@ -134,14 +134,16 @@ StatusWith<std::vector<BSONObj>> selectChunkSplitPoints(OperationContext* opCtx,
                                                         const NamespaceString& nss,
                                                         const ShardKeyPattern& shardKeyPattern,
                                                         const ChunkRange& chunkRange,
-                                                        long long chunkSizeBytes) {
+                                                        long long chunkSizeBytes,
+                                                        boost::optional<int> limit) {
     auto shardStatus = Grid::get(opCtx)->shardRegistry()->getShard(opCtx, shardId);
     if (!shardStatus.isOK()) {
         return shardStatus.getStatus();
     }
 
-    const AutoSplitVectorRequest req(
+    AutoSplitVectorRequest req(
         nss, shardKeyPattern.toBSON(), chunkRange.getMin(), chunkRange.getMax(), chunkSizeBytes);
+    req.setLimit(limit);
 
     auto cmdStatus = shardStatus.getValue()->runCommandWithFixedRetryAttempts(
         opCtx,
diff --git a/src/mongo/s/shard_util.h b/src/mongo/s/shard_util.h
index 7558906ab27..0ab14da2811 100644
--- a/src/mongo/s/shard_util.h
+++ b/src/mongo/s/shard_util.h
@@ -78,20 +78,20 @@ StatusWith<long long> retrieveCollectionShardSize(OperationContext* opCtx,
 /**
  * Ask the specified shard to figure out the split points for a given chunk.
  *
- * shardId The shard id to query.
- * nss Namespace, which owns the chunk.
- * shardKeyPattern The shard key which corresponds to this sharded namespace.
- * chunkRange Bounds of the chunk to be split.
- * chunkSize Chunk size to target in bytes.
- * maxObjs Limits the number of objects in each chunk. Zero means max, unspecified means use the
- *         server default.
+ * - shardId: the shard id to query.
+ * - nss: namespace, which owns the chunk.
+ * - shardKeyPattern: the shard key which corresponds to this sharded namespace.
+ * - chunkRange: bounds of the chunk to search for split points on.
+ * - chunkSizeBytes: chunk size to target in bytes.
+ * - limit: limits the number of split points to search. Unspecified means no limit
  */
 StatusWith<std::vector<BSONObj>> selectChunkSplitPoints(OperationContext* opCtx,
                                                         const ShardId& shardId,
                                                         const NamespaceString& nss,
                                                         const ShardKeyPattern& shardKeyPattern,
                                                         const ChunkRange& chunkRange,
-                                                        long long chunkSizeBytes);
+                                                        long long chunkSizeBytes,
+                                                        boost::optional<int> limit = boost::none);
 
 /**
  * Asks the specified shard to split the chunk described by min/maxKey into the respective split