fix(broker/bam): downtimes on kpi can be more than one

* enh(tests/bam): test on bam and ignored downtimes on kpi implemented
* chore(doc): CHANGELOG updated
* doc(tests): README updated

REFS: MON-14091

CHANGELOG.md

@@ -26,6 +26,10 @@ A gRPC stream connector did not stop correctly on cbd stop.
 On BAM misconfiguration, cbd could crash. This is fixed now. That was due to
 an issue in mysql code with promises handling.
 
+In a BA configured to ignore its kpi downtimes, if a kpi represented by a
+service has two overlapping downtimes applied. When the first one is cancelled,
+it is as if all the downtimes are removed. This is fixed with this new version.
+
 *Debian*
 
 Default configuration files were not installed on a Debian fresh install.

broker/bam/src/kpi_service.cc

@@ -302,21 +302,25 @@ void kpi_service::service_update(std::shared_ptr<neb::downtime> const& dt,
                                  io::stream* visitor) {
   assert(dt && dt->host_id == _host_id && dt->service_id == _service_id);
   // Update information.
-  _downtimed = dt->was_started && dt->actual_end_time.is_null();
+  bool downtimed = dt->was_started && dt->actual_end_time.is_null();
+  if (!_downtimed && downtimed)
+    _downtimed = true;
+
   if (_downtime_ids.contains(dt->internal_id) && !dt->was_cancelled) {
     log_v2::bam()->trace("Downtime {} already handled in this kpi service",
                          dt->internal_id);
     return;
   }
 
-  if (_downtimed) {
+  if (downtimed) {
     log_v2::bam()->trace("adding in kpi service the impacting downtime {}",
                          dt->internal_id);
     _downtime_ids.insert(dt->internal_id);
   } else {
     log_v2::bam()->trace("removing from kpi service the impacting downtime {}",
                          dt->internal_id);
     _downtime_ids.erase(dt->internal_id);
+    _downtimed = !_downtime_ids.empty();
   }
 
   if (!_event || _event->in_downtime != _downtimed) {

tests/bam/inherited_downtime.robot

@@ -127,3 +127,85 @@ BEBAMIDT2
 
 	Stop Engine
 	Stop Broker
+
+BEBAMIGNDT1
+	[Documentation]	A BA of type 'worst' with two services is configured. The downtime policy on this ba is "Ignore the indicator in the calculation". The BA is in critical state, because of the second critical service. Then we apply two downtimes on this last one. The BA state is ok because of the policy on indicators. A first downtime is cancelled, the BA is still OK, but when the second downtime is cancelled, the BA should be CRITICAL.
+	[Tags]	broker	downtime	engine	bam
+	Clear Commands Status
+	Config Broker	module
+	Config Broker	central
+	Broker Config Log	central	bam	trace
+	Config Broker	rrd
+	Config Engine	${1}
+
+	Clone Engine Config To DB
+	Add Bam Config To Engine
+
+	@{svc}=	Set Variable	${{ [("host_16", "service_313"), ("host_16", "service_314")] }}
+	Create BA With Services	test	worst	${svc}  ignore
+	Add Bam Config To Broker	central
+	# Command of service_314 is set to critical
+	${cmd_1}=	Get Command Id	313
+	Log To Console	service_314 has command id ${cmd_1}
+	Set Command Status	${cmd_1}	0
+	${cmd_2}=	Get Command Id	314
+	Log To Console	service_314 has command id ${cmd_2}
+	Set Command Status	${cmd_2}	2
+	Start Broker
+	Start Engine
+	Sleep	5s
+
+	# KPI set to ok
+	Repeat Keyword	3 times	Process Service Check Result	host_16	service_313	0	output critical for 313
+	${result}=	Check Service Status With Timeout	host_16	service_313	0	60
+	Should Be True	${result}	msg=The service (host_16,service_313) is not OK as expected
+
+	# KPI set to critical
+	Repeat Keyword	3 times	Process Service Check Result	host_16	service_314	2	output critical for 314
+	${result}=	Check Service Status With Timeout	host_16	service_314	2	60
+	Should Be True	${result}	msg=The service (host_16,service_314) is not CRITICAL as expected
+
+	# The BA should become critical
+	${result}=	Check Ba Status With Timeout	test	2	60
+	Should Be True	${result}	msg=The BA ba_1 is not CRITICAL as expected
+	Log To console	The BA is critical.
+
+	# Two downtimes are applied on service_314
+	Schedule Service Downtime	host_16	service_314	3600
+	${result}=	Check Service Downtime With Timeout	host_16	service_314	1	60
+	Should Be True	${result}	msg=The service (host_16, service_314) is not in downtime as it should be
+	Log to console	One downtime applied to service_314.
+
+	Schedule Service Downtime	host_16	service_314	1800
+	${result}=	Check Service Downtime With Timeout	host_16	service_314	2	60
+	Should Be True	${result}	msg=The service (host_16, service_314) is not in downtime as it should be
+	Log to console	Two downtimes applied to service_314.
+
+	${result}=	Check Service Downtime With Timeout	_Module_BAM_1	ba_1	0	60
+	Should Be True	${result}	msg=The BA ba_1 is in downtime but should not
+	Log to console	The BA is configured to ignore kpis in downtime
+
+	${result}=	Check Ba Status With Timeout	test	0	60
+	Should Be True	${result}	msg=The service in downtime should be ignored while computing the state of this BA.
+	Log to console	The BA is OK, since the critical service is in downtime.
+
+	# The first downtime is deleted
+	Delete Service Downtime	host_16	service_314
+
+	${result}=	Check Service Downtime With Timeout	host_16	service_314	1	60
+	Should Be True	${result}	msg=The service (host_16, service_314) does not contain 1 downtime as it should
+	Log to console	Still one downtime applied to service_314.
+
+	${result}=	Check Ba Status With Timeout	test	0	60
+	Should Be True	${result}	msg=The BA is not OK whereas the service_314 is still in downtime.
+	Log to console	The BA is still OK
+
+	# The second downtime is deleted
+	Delete Service Downtime	host_16	service_314
+	${result}=	Check Ba Status With Timeout	test	2	60
+	Should Be True	${result}	msg=The critical service is no more in downtime, the BA should be critical.
+	Log to console	The BA is now critical (no more downtime)
+
+	Stop Engine
+	Kindly Stop Broker
+

tests/resources/Common.py

@@ -340,7 +340,7 @@ def check_service_status_with_timeout(hostname: str, service_desc: str, status:
                 cursor.execute("SELECT s.state FROM services s LEFT JOIN hosts h ON s.host_id=h.host_id WHERE s.description=\"{}\" AND h.name=\"{}\"".format(
                     service_desc, hostname))
                 result = cursor.fetchall()
-                if result[0]['state'] and int(result[0]['state']) == status:
+                if result[0]['state'] is not None and int(result[0]['state']) == int(status):
                     return True
         time.sleep(5)
     return False
@@ -446,7 +446,7 @@ def check_ba_status_with_timeout(ba_name: str, status: int, timeout: int):
                 cursor.execute(
                     "SELECT current_status FROM mod_bam WHERE name='{}'".format(ba_name))
                 result = cursor.fetchall()
-                if result[0]['current_status'] and int(result[0]['current_status']) == status:
+                if result[0]['current_status'] is not None and int(result[0]['current_status']) == status:
                     return True
         time.sleep(5)
     return False
@@ -484,7 +484,7 @@ def delete_service_downtime(hst: str, svc: str):
 
     with connection:
         with connection.cursor() as cursor:
-            cursor.execute("select d.internal_id from downtimes d inner join hosts h on d.host_id=h.host_id inner join services s on d.service_id=s.service_id where d.cancelled='0' and s.scheduled_downtime_depth='1' and s.description='{}' and h.name='{}'".format(svc, hst))
+            cursor.execute("select d.internal_id from downtimes d inner join hosts h on d.host_id=h.host_id inner join services s on d.service_id=s.service_id where d.cancelled='0' and s.scheduled_downtime_depth<>'0' and s.description='{}' and h.name='{}' LIMIT 1".format(svc, hst))
             result = cursor.fetchall()
             did = int(result[0]['internal_id'])
 

tests/resources/Engine.py

@@ -636,9 +636,9 @@ def add_bam_config_to_engine():
     dbconf.init_bam()
 
 
-def create_ba_with_services(name: str, typ: str, svc: list):
+def create_ba_with_services(name: str, typ: str, svc: list, dt_policy="inherit"):
     global dbconf
-    dbconf.create_ba_with_services(name, typ, svc)
+    dbconf.create_ba_with_services(name, typ, svc, dt_policy)
 
 
 def get_command_id(service: int):

tests/resources/db_conf.py

@@ -170,7 +170,7 @@ def create_conf_db(self):
                         hid += 1
                     connection.commit()
 
-    def create_ba_with_services(self, name:str, typ:str, svc:[(str,str)]):
+    def create_ba_with_services(self, name:str, typ:str, svc:[(str,str)], dt_policy):
         connection = pymysql.connect(host=DB_HOST,
                                      user=DB_USER,
                                      password=DB_PASS,
@@ -184,7 +184,14 @@ def create_ba_with_services(self, name:str, typ:str, svc:[(str,str)]):
             elif typ == 'worst':
                 t = 2
             with connection.cursor() as cursor:
-                cursor.execute("INSERT INTO mod_bam (name, state_source, activate,id_reporting_period,level_w,level_c,id_notification_period,notifications_enabled,event_handler_enabled, inherit_kpi_downtimes) VALUES ('{}',{},'1',1, 80, 70, 1,'0', '0','1')".format(name, t))
+                if dt_policy == "inherit":
+                    inherit_dt = 1
+                elif dt_policy == "ignore":
+                    inherit_dt = 2
+                else:
+                    inherit_dt = 0
+
+                cursor.execute("INSERT INTO mod_bam (name, state_source, activate,id_reporting_period,level_w,level_c,id_notification_period,notifications_enabled,event_handler_enabled, inherit_kpi_downtimes) VALUES ('{}',{},'1',1, 80, 70, 1,'0', '0','{}')".format(name, t, inherit_dt))
                 id_ba = cursor.lastrowid
                 sid = self.engine.create_bam_service("ba_{}".format(id_ba), name, "_Module_BAM_1", "centreon-bam-check!{}".format(id_ba))
                 cursor.execute("INSERT INTO service (service_id, service_description, display_name, service_active_checks_enabled, service_passive_checks_enabled,service_register) VALUES ({0}, \"ba_{1}\",\"{2}\",'2','2','2')".format(sid, id_ba, name))