Add --skip-acls option for exporting.

dbclient/ClustersClient.py

@@ -607,23 +607,26 @@ def log_cluster_configs(self, log_file='clusters.log', acl_log_file='acl_cluster
                     cluster_json['aws_attributes'] = aws_conf
                 cluster_perms = self.get_cluster_acls(cluster_json['cluster_id'], cluster_json['cluster_name'])
 
+                args = parser.get_pipeline_parser().parse_args()
+
                 if users_list:
                     acls = [acl for acl in cluster_perms.get("access_control_list") if
                             (acl.get("group_name", "") in self.groups_to_keep) or
                             (acl.get("user_name", "") in users_list) or
                             (acl.get("group_name", "") == "users")]
                     cluster_perms["access_control_list"] = acls
 
-                    if cluster_perms['http_status_code'] == 200 and acls:
+                    if not args.skip_acls:
+                        if cluster_perms['http_status_code'] == 200 and acls:
+                            acl_log_fp.write(json.dumps(cluster_perms) + '\n')
+                        else:
+                            error_logger.error(f'Failed to get cluster ACL: {cluster_perms}')
+                elif not args.skip_acls:
+                    if cluster_perms['http_status_code'] == 200:
                         acl_log_fp.write(json.dumps(cluster_perms) + '\n')
                     else:
                         error_logger.error(f'Failed to get cluster ACL: {cluster_perms}')
 
-                elif cluster_perms['http_status_code'] == 200:
-                    acl_log_fp.write(json.dumps(cluster_perms) + '\n')
-                else:
-                    error_logger.error(f'Failed to get cluster ACL: {cluster_perms}')
-
                 if filter_user:
                     if cluster_json['creator_user_name'] == filter_user:
                         log_fp.write(json.dumps(cluster_json) + '\n')

dbclient/JobsClient.py

@@ -123,6 +123,12 @@ def log_job_configs(self, users_list=None, groups_list = None, log_file='jobs.lo
                 # reset the original struct with the new settings
                 x['settings'] = job_settings
 
+                args = parser.get_pipeline_parser().parse_args()
+
+                if args.skip_acls:
+                    log_fp.write(json.dumps(x) + '\n')
+                    continue
+
                 # get ACLs and check that the job has one owner before writing
                 job_perms = self.get(f'/preview/permissions/jobs/{job_id}')
                 if not logging_utils.log_response_error(error_logger, job_perms):

dbclient/parser.py

@@ -551,6 +551,9 @@ def get_pipeline_parser() -> argparse.ArgumentParser:
     parser.add_argument('--keep-tasks', nargs='+', type=str, action=ValidateSkipTasks, default=[],
                         help='List of tasks to run in the pipeline; overrides --skip-tasks.')
 
+    parser.add_argument('--skip-acls', action='store_true',
+                        help='Skip exporting ACLs whose absence may cause errors.')
+
     parser.add_argument('--num-parallel', type=int, default=4, help='Number of parallel threads to use to '
                                                                           'export/import')
 

migration_pipeline.py

@@ -97,6 +97,12 @@ def build_export_pipeline(client_config, checkpoint_service, args) -> Pipeline:
     else:
         skip_tasks = args.skip_tasks
 
+    if args.skip_acls:
+        skip_tasks.append(wmconstants.WORKSPACE_ACLS)
+        skip_tasks.append(wmconstants.METASTORE_TABLE_ACLS)
+
+        skip_tasks = list(set(skip_tasks))
+
     source_info_file = os.path.join(client_config['export_dir'], "source_info.txt")
     with open(source_info_file, 'w') as f:
         f.write(client_config['url'])