DAOS-16825 test: Support register cleanup for all Test classes

Support calling register cleanup methods for tests based upon the Test
and TestWithoutServers classes. Also remove stopping agents as part of
calling TestWithServers.stop_servers() since DAOS-6873 is no longer an
issue.

Skip-unit-tests: true
Skip-fault-injection-test: true
Test-tag: pr daily_regression full_regression

Required-githooks: true

Signed-off-by: Phil Henderson <[email protected]>

src/tests/ftest/control/config_generate_run.py

@@ -52,6 +52,7 @@ def test_config_generate_run(self):
             control_metadata = os.path.join(self.test_env.log_dir, 'control_metadata')
 
         # Call dmg config generate. AP is always the first server host.
+        self.log_step("Generating server configuration")
         server_host = self.hostlist_servers[0]
         result = self.get_dmg_command().config_generate(
             access_points=server_host, num_engines=num_engines, scm_only=scm_only,
@@ -66,25 +67,20 @@ def test_config_generate_run(self):
         # Stop and restart daos_server. self.start_server_managers() has the
         # server start-up check built into it, so if there's something wrong,
         # it'll throw an error.
-        self.log.info("Stopping servers")
+        self.log_step("Stopping servers")
         self.stop_servers()
 
         # Create a new server config from generated_yaml and update SCM-related
         # data in engine_params so that the cleanup before the server start
         # works.
-        self.log.info("Copy config to %s and update engine_params", self.test_env.server_config)
+        self.log_step(f"Copy config to {self.test_env.server_config} and update engine_params")
         self.server_managers[0].update_config_file_from_file(generated_yaml)
 
         # Start server with the generated config.
-        self.log.info("Restarting server with the generated config")
+        self.log_step("Restarting server with the generated config")
         try:
-            agent_force = self.start_server_managers(force=True)
+            self.start_server_managers(force=True)
         except ServerFailed as error:
             self.fail(f"Restarting server failed! {error}")
 
-        # We don't need agent for this test. However, when we stop the server,
-        # agent is also stopped. Then the harness checks that the agent is
-        # running during the teardown. If agent isn't running at that point, it
-        # would cause an error, so start it here.
-        self.log.info("Restarting agents")
-        self.start_agent_managers(force=agent_force)
+        self.log.info("Test passed")

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

@@ -476,6 +476,9 @@ def tearDown(self):
         self.report_timeout()
         super().tearDown()
 
+        # Execute any tear down steps in the reverse order of which they were registered.
+        self._teardown_errors.extend(self._cleanup())
+
         # Clean up any temporary files
         self._teardown_errors.extend(self.remove_temp_test_dir())
 
@@ -894,6 +897,7 @@ def start_agents(self, agent_groups=None, force=False):
         self.setup_agents(agent_groups)
         if self.agent_managers:
             self.start_agent_managers(force)
+            self.register_cleanup(self.stop_agents)
 
     def start_servers(self, server_groups=None, force=False):
         """Start the daos_server processes.
@@ -917,6 +921,7 @@ def start_servers(self, server_groups=None, force=False):
         self.setup_servers(server_groups)
         if self.server_managers:
             force_agent_start = self.start_server_managers(force)
+            self.register_cleanup(self.stop_servers)
         return force_agent_start
 
     def restart_servers(self):
@@ -1396,24 +1401,14 @@ def tearDown(self):
         # class (see DAOS-1452/DAOS-9941 and Avocado issue #5217 with
         # associated PR-5224)
         if self.status is not None and self.status != 'PASS' and self.status != 'SKIP':
-            self.__dump_engines_stacks("Test status is {}".format(self.status))
+            self.__dump_engines_stacks(f"Test status is {self.status}")
 
         # Report whether or not the timeout has expired
         self.report_timeout()
 
         # Tear down any test-specific items
         self._teardown_errors = self.pre_tear_down()
 
-        # Destroy any job managers, containers, pools, and dfuse instances next
-        # Eventually this call will encompass all teardown steps
-        self._teardown_errors.extend(self._cleanup())
-
-        # Stop the agents
-        self._teardown_errors.extend(self.stop_agents())
-
-        # Stop the servers
-        self._teardown_errors.extend(self.stop_servers())
-
         super().tearDown()
 
     def pre_tear_down(self):
@@ -1613,11 +1608,6 @@ def stop_servers(self):
                 "Stopping %s group(s) of servers", len(self.server_managers))
             errors.extend(self._stop_managers(self.server_managers, "servers"))
 
-            # Stopping agents whenever servers are stopped for DAOS-6873
-            self.log.info(
-                "Workaround for DAOS-6873: Stopping %s group(s) of agents",
-                len(self.agent_managers))
-            errors.extend(self._stop_managers(self.agent_managers, "agents"))
         return errors
 
     def _stop_managers(self, managers, name):