DAOS-16072 test: Support non-root user client commands

src/tests/ftest/harness/advanced.py

@@ -1,5 +1,5 @@
 """
-  (C) Copyright 2021-2023 Intel Corporation.
+  (C) Copyright 2021-2024 Intel Corporation.
 
   SPDX-License-Identifier: BSD-2-Clause-Patent
 """
@@ -102,11 +102,11 @@ def test_launch_failures(self):
         host = NodeSet(choice(self.server_managers[0].hosts))   # nosec
         self.log.info("Creating launch.py failure trigger files on %s", host)
         failure_trigger = "00_trigger-launch-failure_00"
-        failure_trigger_dir = os.path.join(self.base_test_dir, failure_trigger)
+        failure_trigger_dir = os.path.join(self.test_env.log_dir, failure_trigger)
         failure_trigger_files = [
-            os.path.join(self.base_test_dir, "{}_local.yaml".format(failure_trigger)),
+            os.path.join(self.test_env.log_dir, "{}_local.yaml".format(failure_trigger)),
             os.path.join(os.sep, "etc", "daos", "daos_{}.yml".format(failure_trigger)),
-            os.path.join(self.base_test_dir, "{}.log".format(failure_trigger)),
+            os.path.join(self.test_env.log_dir, "{}.log".format(failure_trigger)),
             os.path.join(failure_trigger_dir, "{}.log".format(failure_trigger)),
             os.path.join(os.sep, "tmp", "daos_dump_{}.txt".format(failure_trigger)),
             os.path.join(self.tmp, "valgrind_{}".format(failure_trigger)),

src/tests/ftest/launch.py

@@ -280,7 +280,7 @@ def _run(self, args):
             else:
                 set_test_environment(
                     logger, test_env, args.test_servers, args.test_clients, args.provider,
-                    args.insecure_mode, self.details)
+                    args.insecure_mode, self.details, args.agent_user, args.test_log_dir)
         except TestEnvironmentException as error:
             message = f"Error setting up test environment: {str(error)}"
             return self.get_exit_status(1, message, "Setup", sys.exc_info())
@@ -318,12 +318,13 @@ def _run(self, args):
             return self.get_exit_status(0, "Listing tests complete")
 
         # Setup the fuse configuration
-        try:
-            setup_fuse_config(logger, args.test_servers | args.test_clients)
-        except LaunchException:
-            # Warn but don't fail
-            message = "Issue detected setting up the fuse configuration"
-            setup_result.warn_test(logger, "Setup", message, sys.exc_info())
+        if args.fuse_setup:
+            try:
+                setup_fuse_config(logger, args.test_servers | args.test_clients)
+            except LaunchException:
+                # Warn but don't fail
+                message = "Issue detected setting up the fuse configuration"
+                setup_result.warn_test(logger, "Setup", message, sys.exc_info())
 
         # Get the core file pattern information
         core_files = {}
@@ -347,8 +348,8 @@ def _run(self, args):
             group.update_test_yaml(
                 logger, args.scm_size, args.scm_mount, args.extra_yaml,
                 args.timeout_multiplier, args.override, args.verbose, args.include_localhost)
-        except (RunException, YamlException) as e:
-            message = "Error modifying the test yaml files: {}".format(e)
+        except (RunException, YamlException) as error:
+            message = f"Error modifying the test yaml files: {str(error)}"
             status |= self.get_exit_status(1, message, "Setup", sys.exc_info())
         except StorageException:
             message = "Error detecting storage information for test yaml files"
@@ -529,6 +530,12 @@ def main():
         "-a", "--archive",
         action="store_true",
         help="archive host log files in the avocado job-results directory")
+    parser.add_argument(
+        "-au", "--agent_user",
+        action="store",
+        default=None,
+        type=str,
+        help="user account to use when running the daos_agent")
     parser.add_argument(
         "-c", "--clear_mounts",
         action="append",
@@ -551,6 +558,10 @@ def main():
         "--failfast",
         action="store_true",
         help="stop the test suite after the first failure")
+    parser.add_argument(
+        "-fs", "--fuse_setup",
+        action="store_true",
+        help="enable setting up fuse configuration files")
     parser.add_argument(
         "-i", "--include_localhost",
         action="store_true",
@@ -573,7 +584,7 @@ def main():
         help="modify the test yaml files but do not run the tests")
     parser.add_argument(
         "-mo", "--mode",
-        choices=['normal', 'manual', 'ci'],
+        choices=['normal', 'manual', 'ci', 'agent_user'],
         default='normal',
         help="provide the mode of test to be run under. Default is normal, "
              "in which the final return code of launch.py is still zero if "
@@ -670,6 +681,12 @@ def main():
         default=NodeSet(),
         help="comma-separated list of hosts to use as replacement values for "
              "client placeholders in each test's yaml file")
+    parser.add_argument(
+        "-tld", "--test_log_dir",
+        action="store",
+        default=None,
+        type=str,
+        help="test log directory base path")
     parser.add_argument(
         "-th", "--logs_threshold",
         action="store",
@@ -733,10 +750,26 @@ def main():
         args.slurm_install = True
         args.slurm_setup = True
         args.user_create = True
+        args.fuse_setup = True
         args.clear_mounts.append("/mnt/daos")
         args.clear_mounts.append("/mnt/daos0")
         args.clear_mounts.append("/mnt/daos1")
 
+    elif args.mode == "agent_user":
+        if args.agent_user is None:
+            # Run the agent with the current user by default
+            args.agent_user = getpass.getuser()
+        if os.environ.get("DAOS_TEST_LOG_DIR", args.test_log_dir) is None:
+            # Use a user-specific test log dir by default
+            args.test_log_dir = os.path.join(
+                os.sep, "var", "tmp", f"daos_testing_{args.agent_user}")
+        args.process_cores = False
+        args.logs_threshold = None
+        args.slurm_install = False
+        args.slurm_setup = False
+        args.user_create = False
+        args.fuse_setup = False
+
     # Setup the Launch object
     launch = Launch(args.name, args.mode, args.slurm_install, args.slurm_setup)
 

src/tests/ftest/util/agent_utils.py

@@ -1,12 +1,11 @@
 """
-  (C) Copyright 2019-2023 Intel Corporation.
+  (C) Copyright 2019-2024 Intel Corporation.
 
   SPDX-License-Identifier: BSD-2-Clause-Patent
 """
 import os
 import re
 import socket
-from getpass import getuser
 
 from agent_utils_params import DaosAgentTransportCredentials, DaosAgentYamlParameters
 from ClusterShell.NodeSet import NodeSet
@@ -65,7 +64,7 @@ def get_agent_command(group, cert_dir, bin_dir, config_file, config_temp=None):
 class DaosAgentCommand(YamlCommand):
     """Defines an object representing a daos_agent command."""
 
-    def __init__(self, path="", yaml_cfg=None, timeout=15):
+    def __init__(self, path="", yaml_cfg=None, timeout=15, run_user=None):
         """Create a daos_agent command object.
 
         Args:
@@ -74,10 +73,12 @@ def __init__(self, path="", yaml_cfg=None, timeout=15):
                 parameters. Defaults to None.
             timeout (int, optional): number of seconds to wait for patterns to
                 appear in the subprocess output. Defaults to 60 seconds.
+            run_user (str, optional): user to run as. Defaults to None, which uses the current user.
         """
         super().__init__(
-            "/run/agent_config/*", "daos_agent", path, yaml_cfg, timeout)
+            "/run/agent_config/*", "daos_agent", path, yaml_cfg, timeout, run_user)
         self.pattern = "listening on "
+        self.run_user = run_user
 
         # If specified use the configuration file from the YamlParameters object
         default_yaml_file = None
@@ -206,7 +207,7 @@ class DaosAgentManager(SubprocessManager):
     """Manages the daos_agent execution on one or more hosts."""
 
     def __init__(self, group, bin_dir, cert_dir, config_file, config_temp=None,
-                 manager="Orterun", outputdir=None):
+                 manager="Orterun", outputdir=None, run_user=None):
         """Initialize a DaosAgentManager object.
 
         Args:
@@ -223,8 +224,8 @@ def __init__(self, group, bin_dir, cert_dir, config_file, config_temp=None,
             outputdir (str, optional): path to avocado test outputdir. Defaults
                 to None.
         """
-        agent_command = get_agent_command(
-            group, cert_dir, bin_dir, config_file, config_temp)
+        agent_command = get_agent_command(group, cert_dir, bin_dir, config_file, config_temp)
+        agent_command.run_user = 'root'
         super().__init__(agent_command, manager)
 
         # Set the correct certificate file ownership
@@ -265,11 +266,10 @@ def start(self):
             self._hosts, self.manager.command)
 
         # Copy certificates
-        self.manager.job.copy_certificates(
-            get_log_file("daosCA/certs"), self._hosts)
+        self.manager.job.copy_certificates(get_log_file("daosCA/certs"), self._hosts)
 
         # Verify the socket directory exists when using a non-systemctl manager
-        self.verify_socket_directory(getuser())
+        self.verify_socket_directory(self.manager.job.certificate_owner)
 
         super().start()
 

src/tests/ftest/util/apricot/apricot/test.py

@@ -68,8 +68,8 @@ def __init__(self, *args, **kwargs):
         self.test_id = self.get_test_name()
 
         # Define a test unique temporary directory
-        self.base_test_dir = os.getenv("DAOS_TEST_LOG_DIR", "/tmp")
-        self.test_dir = os.path.join(self.base_test_dir, self.test_id)
+        self.test_env = TestEnvironment()
+        self.test_dir = os.path.join(self.test_env.log_dir, self.test_id)
         if not os.path.exists(self.test_dir):
             os.makedirs(self.test_dir)
 
@@ -545,7 +545,8 @@ def tearDown(self):
         """Tear down after each test case."""
         self.report_timeout()
         if self.fault_injection:
-            self._teardown_errors.extend(self.fault_injection.stop())
+            self._teardown_errors.extend(
+                self.fault_injection.stop(self.test_env.agent_user or "root"))
         super().tearDown()
 
     def stop_leftover_processes(self, processes, hosts):
@@ -744,12 +745,6 @@ def setUp(self):
         self.__dump_engine_ult_on_failure = self.params.get(
             "dump_engine_ult_on_failure", "/run/setup/*", True)
 
-        # # Find a configuration that meets the test requirements
-        # self.config = Configuration(
-        #     self.params, self.hostlist_servers, debug=self.debug)
-        # if not self.config.set_config(self):
-        #     self.cancel("Test requirements not met!")
-
         # Create host files - In the future this should be the responsibility of
         # tests/classes that need a host file and hostfile_clients should not be
         # a property of this class.
@@ -769,8 +764,8 @@ def setUp(self):
             hosts.add(self.hostlist_clients)
         # Copy the fault injection files to the hosts.
         self.fault_injection.copy_fault_files(hosts)
-        lines = get_file_listing(hosts, self.test_dir).stdout_text.splitlines()
-        for line in lines:
+        listing = get_file_listing(hosts, self.test_dir, self.test_env.agent_user or "root")
+        for line in listing.stdout_text.splitlines():
             self.log.debug("  %s", line)
 
         if not self.start_servers_once or self.name.uid == 1:
@@ -1117,8 +1112,7 @@ def add_server_manager(self, group=None, svr_config_file=None,
             group = self.server_group
         if svr_config_file is None and self.server_manager_class == "Systemctl":
             svr_config_file = get_default_config_file("server")
-            svr_config_temp = self.get_config_file(
-                group, "server", self.test_dir)
+            svr_config_temp = self.get_config_file(group, "server", self.test_dir)
         elif svr_config_file is None:
             svr_config_file = self.get_config_file(group, "server")
             svr_config_temp = None
@@ -1345,7 +1339,8 @@ def remove_temp_test_dir(self):
             "Removing temporary test files in %s from %s",
             self.test_dir, str(NodeSet.fromlist(all_hosts)))
         result = run_remote(
-            self.log, all_hosts, command_as_user("rm -fr {}".format(self.test_dir), "root"))
+            self.log, all_hosts,
+            command_as_user("rm -fr {}".format(self.test_dir), self.test_env.agent_user or "root"))
         if not result.passed:
             errors.append("Error removing temporary test files on {}".format(result.failed_hosts))
         return errors