1 files changed, 74 insertions, 0 deletions
diff --git a/src/ceph/qa/tasks/mgr/test_module_selftest.py b/src/ceph/qa/tasks/mgr/test_module_selftest.py
new file mode 100644
index 0000000..2776fb8
--- /dev/null
+++ b/src/ceph/qa/tasks/mgr/test_module_selftest.py
@@ -0,0 +1,74 @@
+
+import time
+import requests
+
+from tasks.mgr.mgr_test_case import MgrTestCase
+
+
+class TestModuleSelftest(MgrTestCase):
+    """
+    That modules with a self-test command can be loaded and execute it
+    without errors.
+
+    This is not a substitute for really testing the modules, but it
+    is quick and is designed to catch regressions that could occur
+    if data structures change in a way that breaks how the modules
+    touch them.
+    """
+    MGRS_REQUIRED = 1
+
+    def _selftest_plugin(self, module_name):
+        self._load_module(module_name)
+
+        # Execute the module's self-test routine
+        self.mgr_cluster.mon_manager.raw_cluster_cmd(module_name, "self-test")
+
+    def test_zabbix(self):
+        self._selftest_plugin("zabbix")
+
+    def test_prometheus(self):
+        self._selftest_plugin("prometheus")
+
+    def test_influx(self):
+        self._selftest_plugin("influx")
+
+    def test_selftest_run(self):
+        self._load_module("selftest")
+        self.mgr_cluster.mon_manager.raw_cluster_cmd("mgr", "self-test", "run")
+
+    def test_selftest_command_spam(self):
+        # Use the selftest module to stress the mgr daemon
+        self._load_module("selftest")
+
+        # Use the dashboard to test that the mgr is still able to do its job
+        self._assign_ports("dashboard", "server_port")
+        self._load_module("dashboard")
+
+        original_active = self.mgr_cluster.get_active_id()
+        original_standbys = self.mgr_cluster.get_standby_ids()
+
+        self.mgr_cluster.mon_manager.raw_cluster_cmd("mgr", "self-test",
+                                                     "background", "start",
+                                                     "command_spam")
+
+        dashboard_uri = self._get_uri("dashboard")
+
+        delay = 10
+        periods = 10
+        for i in range(0, periods):
+            t1 = time.time()
+            # Check that an HTTP module remains responsive
+            r = requests.get(dashboard_uri)
+            self.assertEqual(r.status_code, 200)
+
+            # Check that a native non-module command remains responsive
+            self.mgr_cluster.mon_manager.raw_cluster_cmd("osd", "df")
+
+            time.sleep(delay - (time.time() - t1))
+
+        self.mgr_cluster.mon_manager.raw_cluster_cmd("mgr", "self-test",
+                                                     "background", "stop")
+
+        # Check that all mgr daemons are still running
+        self.assertEqual(original_active, self.mgr_cluster.get_active_id())
+        self.assertEqual(original_standbys, self.mgr_cluster.get_standby_ids())