[enhancement](cloud) retry TXN_MAYBE_COMMITTED and add UT (#59563)

Author: luwei16

cloud/src/meta-service/meta_service.h

@@ -1017,6 +1017,7 @@ class MetaServiceProxy final : public MetaService {
             if (code != MetaServiceCode::KV_TXN_STORE_GET_RETRYABLE &&
                 code != MetaServiceCode::KV_TXN_STORE_COMMIT_RETRYABLE &&
                 code != MetaServiceCode::KV_TXN_STORE_CREATE_RETRYABLE &&
+                code != MetaServiceCode::KV_TXN_MAYBE_COMMITTED &&
                 code != MetaServiceCode::KV_TXN_TOO_OLD &&
                 (!config::enable_retry_txn_conflict || code != MetaServiceCode::KV_TXN_CONFLICT)) {
                 return;
@@ -1040,6 +1041,8 @@ class MetaServiceProxy final : public MetaService {
                         code == MetaServiceCode::KV_TXN_STORE_COMMIT_RETRYABLE   ? KV_TXN_COMMIT_ERR
                         : code == MetaServiceCode::KV_TXN_STORE_GET_RETRYABLE    ? KV_TXN_GET_ERR
                         : code == MetaServiceCode::KV_TXN_STORE_CREATE_RETRYABLE ? KV_TXN_CREATE_ERR
+                        : code == MetaServiceCode::KV_TXN_MAYBE_COMMITTED
+                                ? MetaServiceCode::KV_TXN_MAYBE_COMMITTED
                         : code == MetaServiceCode::KV_TXN_CONFLICT
                                 ? KV_TXN_CONFLICT_RETRY_EXCEEDED_MAX_TIMES
                                 : MetaServiceCode::KV_TXN_TOO_OLD);

cloud/src/meta-service/meta_service_helper.h

@@ -252,8 +252,9 @@ inline MetaServiceCode cast_as(TxnErrorCode code) {
             }
         }
         [[fallthrough]];
-    case TxnErrorCode::TXN_KEY_NOT_FOUND:
     case TxnErrorCode::TXN_MAYBE_COMMITTED:
+        return MetaServiceCode::KV_TXN_MAYBE_COMMITTED;
+    case TxnErrorCode::TXN_KEY_NOT_FOUND:
     case TxnErrorCode::TXN_TIMEOUT:
     case TxnErrorCode::TXN_INVALID_ARGUMENT:
     case TxnErrorCode::TXN_INVALID_DATA:

cloud/test/meta_service_test.cpp

@@ -8266,6 +8266,50 @@ TEST(MetaServiceTxnStoreRetryableTest, DoNotReturnRetryableCode) {
     config::txn_store_retry_times = retry_times;
 }
 
+TEST(MetaServiceTxnStoreRetryableTest, RetryMaybeCommittedCode) {
+    size_t index = 0;
+    SyncPoint::get_instance()->set_call_back("update_delete_bitmap:commit:err", [&](auto&& args) {
+        ++index;
+        *doris::try_any_cast<TxnErrorCode*>(args[2]) = TxnErrorCode::TXN_MAYBE_COMMITTED;
+    });
+    SyncPoint::get_instance()->enable_processing();
+    int32_t retry_times = config::txn_store_retry_times;
+    bool enable_retry = config::enable_txn_store_retry;
+    int64_t max_txn_commit_byte = config::max_txn_commit_byte;
+    config::txn_store_retry_times = 2;
+    config::enable_txn_store_retry = true;
+    config::max_txn_commit_byte = 1;
+
+    auto service = get_meta_service();
+
+    brpc::Controller cntl;
+    UpdateDeleteBitmapRequest req;
+    UpdateDeleteBitmapResponse resp;
+    req.set_cloud_unique_id("test_cloud_unique_id");
+    req.set_table_id(100);
+    req.set_partition_id(123);
+    req.set_lock_id(-3);
+    req.set_without_lock(true);
+    req.set_initiator(-1);
+    req.set_tablet_id(333);
+    req.add_rowset_ids("r1");
+    req.add_segment_ids(0);
+    req.add_versions(2);
+    req.add_segment_delete_bitmaps("abc");
+
+    service->update_delete_bitmap(&cntl, &req, &resp, nullptr);
+
+    ASSERT_EQ(resp.status().code(), MetaServiceCode::KV_TXN_MAYBE_COMMITTED)
+            << " status is " << resp.status().msg() << ", code=" << resp.status().code();
+    EXPECT_GE(index, static_cast<size_t>(config::txn_store_retry_times + 1));
+
+    SyncPoint::get_instance()->disable_processing();
+    SyncPoint::get_instance()->clear_all_call_backs();
+    config::txn_store_retry_times = retry_times;
+    config::enable_txn_store_retry = enable_retry;
+    config::max_txn_commit_byte = max_txn_commit_byte;
+}
+
 TEST(MetaServiceTest, GetClusterStatusTest) {
     auto meta_service = get_meta_service();
 

gensrc/proto/cloud.proto

@@ -1741,6 +1741,7 @@ enum MetaServiceCode {
     KV_TXN_STORE_COMMIT_RETRYABLE = 1009;
     KV_TXN_STORE_CREATE_RETRYABLE = 1010;
     KV_TXN_TOO_OLD = 1011;
+    KV_TXN_MAYBE_COMMITTED = 1012;
 
     //Doris error
     TXN_GEN_ID_ERR = 2001;