[fix](group commit) Add retry when delete bitmap lock expired on grou…

…p commit (#37600)

group commit writing mow table may encouter delete bitmap lock expired,
add retry mechanism to avoid group commit fail.

be/src/runtime/group_commit_mgr.cpp

@@ -23,6 +23,7 @@
 #include <chrono>
 
 #include "client_cache.h"
+#include "cloud/config.h"
 #include "common/compiler_util.h"
 #include "common/config.h"
 #include "common/status.h"
@@ -446,13 +447,26 @@ Status GroupCommitTable::_finish_group_commit_load(int64_t db_id, int64_t table_
  }
  TLoadTxnCommitResult result;
  TNetworkAddress master_addr = _exec_env->master_info()->network_address;
- st = ThriftRpcHelper::rpc<FrontendServiceClient>(
- master_addr.hostname, master_addr.port,
- [&request, &result](FrontendServiceConnection& client) {
- client->loadTxnCommit(result, request);
- },
- 10000L);
- result_status = Status::create(result.status);
+ int retry_times = 0;
+ while (retry_times < config::mow_stream_load_commit_retry_times) {
+ st = ThriftRpcHelper::rpc<FrontendServiceClient>(
+ master_addr.hostname, master_addr.port,
+ [&request, &result](FrontendServiceConnection& client) {
+ client->loadTxnCommit(result, request);
+ },
+ 10000L);
+ result_status = Status::create(result.status);
+ // DELETE_BITMAP_LOCK_ERROR will be retried
+ if (result_status.ok() || !result_status.is<ErrorCode::DELETE_BITMAP_LOCK_ERROR>()) {
+ break;
+ }
+ LOG_WARNING("Failed to commit txn on group commit")
+ .tag("label", label)
+ .tag("txn_id", txn_id)
+ .tag("retry_times", retry_times)
+ .error(result_status);
+ retry_times++;
+ }
  } else {
  // abort txn
  TLoadTxnRollbackRequest request;