YARN-7848 Force removal of docker containers that do not get removed on first try. Contributed by Eric Yang

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/native/container-executor/impl/container-executor.c

@@ -3206,12 +3206,15 @@ int remove_docker_container(char**argv, int argc) {
   const char *container_id = NULL;
 
   int start_index = 0;
+  char **args = alloc_and_clear_memory(argc + 1, sizeof(char *));
+  args[0] = "-f";
   if (argc == 2) {
     yarn_hierarchy = argv[0];
     container_id = argv[1];
     // Skip the yarn_hierarchy argument for exec_docker_command
     start_index = 1;
   }
+  args[1] = argv[start_index];
 
   pid_t child_pid = fork();
   if (child_pid == -1) {
@@ -3222,15 +3225,15 @@ int remove_docker_container(char**argv, int argc) {
   }
 
   if (child_pid == 0) { // child
-    int rc = exec_docker_command("rm", argv + start_index, argc - start_index);
-    return rc; // Only get here if exec fails
-
+    int rc = exec_docker_command("rm", args, 2);
+    exit_code = rc; // Only get here if exec fails
   } else { // parent
     exit_code = wait_and_get_exit_code(child_pid);
     if (exit_code != 0) {
       exit_code = DOCKER_RUN_FAILED;
     }
   }
+  free(args);
 
   // Clean up cgroups if necessary
   if (yarn_hierarchy != NULL) {

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/native/container-executor/impl/utils/docker-util.c

@@ -722,11 +722,19 @@ int get_docker_rm_command(const char *command_file, const struct configuration *
   }
 
   ret = add_to_args(args, DOCKER_RM_COMMAND);
-  if (ret == 0) {
-    ret = add_to_args(args, container_name);
-    if (ret != 0) {
-      ret = BUFFER_TOO_SMALL;
-    }
+  if (ret != 0) {
+    ret = BUFFER_TOO_SMALL;
+    goto free_and_exit;
+  }
+  ret = add_to_args(args, "-f");
+  if (ret != 0) {
+    ret = BUFFER_TOO_SMALL;
+    goto free_and_exit;
+  }
+  ret = add_to_args(args, container_name);
+  if (ret != 0) {
+    ret = BUFFER_TOO_SMALL;
+    goto free_and_exit;
   }
 free_and_exit:
   free(container_name);

hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/native/container-executor/test/utils/test_docker_util.cc

@@ -304,7 +304,7 @@ namespace ContainerExecutor {
     file_cmd_vec.push_back(
         std::make_pair<std::string, std::string>(
             "[docker-command-execution]\n  docker-command=rm\n  name=container_e1_12312_11111_02_000001",
-            "rm container_e1_12312_11111_02_000001"));
+            "rm -f container_e1_12312_11111_02_000001"));
 
     std::vector<std::pair<std::string, int> > bad_file_cmd_vec;
     bad_file_cmd_vec.push_back(std::make_pair<std::string, int>(
@@ -1711,7 +1711,7 @@ namespace ContainerExecutor {
         "/usr/bin/docker --config=/my-config pull image-id"));
     input_output_map.push_back(std::make_pair<std::string, std::string>(
         "[docker-command-execution]\n  docker-command=rm\n  docker-config=/my-config\n  name=container_e1_12312_11111_02_000001",
-        "/usr/bin/docker --config=/my-config rm container_e1_12312_11111_02_000001"));
+        "/usr/bin/docker --config=/my-config rm -f container_e1_12312_11111_02_000001"));
     input_output_map.push_back(std::make_pair<std::string, std::string>(
         "[docker-command-execution]\n  docker-command=stop\n  docker-config=/my-config\n  name=container_e1_12312_11111_02_000001",
         "/usr/bin/docker --config=/my-config stop container_e1_12312_11111_02_000001"));