From 5f2b95797d8ef052fb74139f40bdd00e411db6ff Mon Sep 17 00:00:00 2001
From: Jia Ke <ke.a.jia@intel.com>
Date: Tue, 17 Dec 2024 20:58:35 +0800
Subject: [PATCH] test

---
 .../hdfs/tests/CMakeLists.txt                 | 48 +++++++++----------
 .../hdfs/tests/HdfsMiniCluster.cpp            | 47 +++++++++++++++---
 2 files changed, 64 insertions(+), 31 deletions(-)

diff --git a/velox/connectors/hive/storage_adapters/hdfs/tests/CMakeLists.txt b/velox/connectors/hive/storage_adapters/hdfs/tests/CMakeLists.txt
index a3d88934805e..c0b32d0fe54c 100644
--- a/velox/connectors/hive/storage_adapters/hdfs/tests/CMakeLists.txt
+++ b/velox/connectors/hive/storage_adapters/hdfs/tests/CMakeLists.txt
@@ -32,27 +32,27 @@ target_link_libraries(
 target_compile_options(velox_hdfs_file_test
                        PRIVATE -Wno-deprecated-declarations)
 
-# add_executable(velox_hdfs_insert_test InsertIntoHdfsTest.cpp
-#                                       HdfsMiniCluster.cpp HdfsUtilTest.cpp)
-
-# add_test(velox_hdfs_insert_test velox_hdfs_insert_test)
-
-# target_link_libraries(
-#   velox_hdfs_insert_test
-#   velox_exec_test_lib
-#   velox_exec
-#   GTest::gtest
-#   GTest::gtest_main
-#   GTest::gmock)
-
-# target_compile_options(velox_hdfs_insert_test
-#                        PRIVATE -Wno-deprecated-declarations)
-
-# # velox_hdfs_insert_test and velox_hdfs_file_test two tests can't run in
-# # parallel due to the port conflict in Hadoop NameNode and DataNode. The
-# # namenode port conflict can be resolved using the -nnport configuration in
-# # hadoop-mapreduce-client-jobclient-3.3.0-tests.jar. However the data node port
-# # cannot be configured. Therefore, we need to make sure that
-# # velox_hdfs_file_test runs only after velox_hdfs_insert_test has finished.
-# set_tests_properties(velox_hdfs_insert_test PROPERTIES DEPENDS
-#                                                        velox_hdfs_file_test)
+add_executable(velox_hdfs_insert_test InsertIntoHdfsTest.cpp
+                                      HdfsMiniCluster.cpp HdfsUtilTest.cpp)
+
+add_test(velox_hdfs_insert_test velox_hdfs_insert_test)
+
+target_link_libraries(
+  velox_hdfs_insert_test
+  velox_exec_test_lib
+  velox_exec
+  GTest::gtest
+  GTest::gtest_main
+  GTest::gmock)
+
+target_compile_options(velox_hdfs_insert_test
+                       PRIVATE -Wno-deprecated-declarations)
+
+# velox_hdfs_insert_test and velox_hdfs_file_test two tests can't run in
+# parallel due to the port conflict in Hadoop NameNode and DataNode. The
+# namenode port conflict can be resolved using the -nnport configuration in
+# hadoop-mapreduce-client-jobclient-3.3.0-tests.jar. However the data node port
+# cannot be configured. Therefore, we need to make sure that
+# velox_hdfs_file_test runs only after velox_hdfs_insert_test has finished.
+set_tests_properties(velox_hdfs_insert_test PROPERTIES DEPENDS
+                                                       velox_hdfs_file_test)
diff --git a/velox/connectors/hive/storage_adapters/hdfs/tests/HdfsMiniCluster.cpp b/velox/connectors/hive/storage_adapters/hdfs/tests/HdfsMiniCluster.cpp
index c262bad3d1d1..4274e994e9c0 100644
--- a/velox/connectors/hive/storage_adapters/hdfs/tests/HdfsMiniCluster.cpp
+++ b/velox/connectors/hive/storage_adapters/hdfs/tests/HdfsMiniCluster.cpp
@@ -14,6 +14,8 @@
  * limitations under the License.
  */
 
+#include <boost/process.hpp>
+
 #include "HdfsMiniCluster.h"
 
 namespace facebook::velox::filesystems::test {
@@ -77,6 +79,14 @@ HdfsMiniCluster::HdfsMiniCluster() {
 }
 
 void HdfsMiniCluster::addFile(std::string source, std::string destination) {
+  std::cout << "Starting to add file from " << source << " to " << destination << std::endl;
+
+  std::ostringstream output;
+  std::ostringstream error;
+
+  boost::process::ipstream outStream;
+  boost::process::ipstream errStream;
+
   auto filePutProcess = std::make_shared<boost::process::child>(
       env_,
       exePath_,
@@ -85,13 +95,36 @@ void HdfsMiniCluster::addFile(std::string source, std::string destination) {
       filesystemUrl,
       filePutOption,
       source,
-      destination);
-  bool isExited =
-      filePutProcess->wait_for(std::chrono::duration<int, std::milli>(5000));
-  if (!isExited) {
-    VELOX_FAIL(
-        "Failed to add file to hdfs, exit code: {}",
-        filePutProcess->exit_code());
+      destination,
+      boost::process::std_out > outStream,
+      boost::process::std_err > errStream);
+
+  std::cout << "Process started with PID: " << filePutProcess->id() << std::endl;
+
+  std::string line;
+  while (outStream && std::getline(outStream, line) && !line.empty())
+      output << line << std::endl;
+
+  while (errStream && std::getline(errStream, line) && !line.empty())
+      error << line << std::endl;
+
+  bool isExited = filePutProcess->wait_for(std::chrono::duration<int, std::milli>(5000));
+
+  if (isExited) {
+    int exitCode = filePutProcess->exit_code();
+    std::cout << "Process exited with code: " << exitCode << std::endl;
+    std::cout << "Process output: " << output.str() << std::endl;
+    std::cout << "Process error: " << error.str() << std::endl;
+
+    if (exitCode != 0) {
+      std::cerr << "Failed to add file to hdfs, exit code: " << exitCode << ", error: " << error.str() << std::endl;
+      throw std::runtime_error("Failed to add file to hdfs, exit code: " + std::to_string(exitCode));
+    }
+  } else {
+    std::cerr << "Failed to add file to hdfs, process did not exit in time" << std::endl;
+    filePutProcess->terminate();
+    std::cerr << "Process terminated" << std::endl;
+    throw std::runtime_error("Failed to add file to hdfs, process did not exit in time");
   }
 }