Fix [#873]: Don't notify if parents down

base/checks.c

@@ -1208,6 +1208,8 @@ int handle_async_service_check_result(service *svc, check_result *cr)
 
 	host * hst                     = NULL;
 
+	int notification_type          = NOTIFICATION_NORMAL;
+
 	log_debug_info(DEBUGL_FUNCTIONS, 0, "handle_async_service_check_result()\n");
 
 	if (svc == NULL) {
@@ -1590,10 +1592,16 @@ int handle_async_service_check_result(service *svc, check_result *cr)
 		handle_event = TRUE;
 	}
 
+	// Recovery notification was not sent
+	if(svc->notified_on != 0 && svc->current_state == STATE_OK && svc->state_type == HARD_STATE) {
+		notification_type = NOTIFICATION_RECOVERY;
+		send_notification = TRUE;
+	}
+
 	if (send_notification == TRUE) {
 
 		/* send notification */
-		if (service_notification(svc, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE) == OK) {
+		if (service_notification(svc, notification_type, NULL, NULL, NOTIFICATION_OPTION_NONE) == OK) {
 
 			/* log state due to notification event when stalking_options N is set */
 			if (should_stalk_notifications(svc)) {
@@ -1602,13 +1610,6 @@ int handle_async_service_check_result(service *svc, check_result *cr)
 		}
 	}
 
-	/* the service recovered, so reset the current notification number and state flags (after the recovery notification has gone out) */
-	/* We don't want to reset notifications if the service is currently flapping because we want recovery notifications */
-	if(svc->current_state == STATE_OK && svc->state_type == HARD_STATE && hard_state_change == TRUE && svc->is_flapping == FALSE) {
-		svc->current_notification_number = 0;
-		svc->notified_on = 0;
-		}
-
 	if (obsess_over_services == TRUE) {
 		obsessive_compulsive_service_check_processor(svc);
 	}
@@ -1634,7 +1635,6 @@ int handle_async_service_check_result(service *svc, check_result *cr)
 
 		/* Reset attempts */
 		if (hard_state_change == TRUE) {
-			svc->current_notification_number = 0;
 			svc->host_problem_at_last_check = FALSE;
 		}
 
@@ -2255,6 +2255,8 @@ int handle_async_host_check_result(host *hst, check_result *cr)
 
 	char * old_plugin_output = NULL;
 
+	int notification_type    = NOTIFICATION_NORMAL;
+
 	log_debug_info(DEBUGL_FUNCTIONS, 0, "handle_async_host_check_result()\n");
 
 	if (is_valid_check_result_data(hst, cr) == FALSE) {
@@ -2482,10 +2484,16 @@ int handle_async_host_check_result(host *hst, check_result *cr)
 		hst->current_attempt = 1;
 	}
 
+	// Recovery notification was not sent
+	if(hst->notified_on != 0 && hst->current_state == HOST_UP && hst->state_type == HARD_STATE) {
+		notification_type = NOTIFICATION_RECOVERY;
+		send_notification = TRUE;
+	}
+
 	if (send_notification == TRUE) {
 
 		/* send notifications */
-		if (host_notification(hst, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE) == OK) {
+		if (host_notification(hst, notification_type, NULL, NULL, NOTIFICATION_OPTION_NONE) == OK) {
 
 			/* log state due to notification event when stalking_options N is set */
 			if (should_stalk_notifications(hst)) {
@@ -2494,13 +2502,6 @@ int handle_async_host_check_result(host *hst, check_result *cr)
 		}
 	}
 
-    /* the host recovered, so reset the current notification number and state flags (after the recovery notification has gone out) */
-	/* We don't want to reset notifications if the host is currently flapping because we want recovery notifications */
-    if(hst->current_state == HOST_UP && hst->state_type == HARD_STATE && hard_state_change == TRUE && hst->is_flapping == FALSE) {
-        hst->current_notification_number = 0;
-        hst->notified_on = 0;
-        }
-
 	if (obsess_over_hosts == TRUE) {
 		obsessive_compulsive_host_check_processor(hst);
 	}

base/commands.c

@@ -5390,10 +5390,7 @@ void clear_host_flapping_state(host *hst) {
 
 		/* should we send a recovery notification? */
 		if (hst->current_state == HOST_UP && hst->check_flapping_recovery_notification == TRUE) {
-			host_notification(hst, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE);
-			/* Similar to what happens when a host recovers in handle_async_host_check_result */
-			hst->current_notification_number = 0;
-			hst->notified_on = 0;
+			host_notification(hst, NOTIFICATION_RECOVERY, NULL, NULL, NOTIFICATION_OPTION_NONE);
 		}
 	}
 
@@ -5450,10 +5447,7 @@ void clear_service_flapping_state(service *svc) {
 
 		/* should we send a recovery notification? */
 		if (svc->current_state == STATE_OK && svc->check_flapping_recovery_notification == TRUE) {
-			service_notification(svc, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE);
-			/* Similar to what happens when a service recovers in handle_async_service_check_result */
-			svc->current_notification_number = 0;
-			svc->notified_on = 0;
+			service_notification(svc, NOTIFICATION_RECOVERY, NULL, NULL, NOTIFICATION_OPTION_NONE);
 		}
 	}
 

base/flapping.c

@@ -375,10 +375,7 @@ void clear_service_flap(service *svc, double percent_change, double high_thresho
 
 		/* should we send a recovery notification? */
 		if(svc->check_flapping_recovery_notification == TRUE && svc->current_state == STATE_OK) {
-			service_notification(svc, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE);
-			/* Similar to what happens when a service recovers in handle_async_service_check_result */
-			svc->current_notification_number = 0;
-			svc->notified_on = 0;
+			service_notification(svc, NOTIFICATION_RECOVERY, NULL, NULL, NOTIFICATION_OPTION_NONE);
 		}
 	}
 
@@ -470,10 +467,7 @@ void clear_host_flap(host *hst, double percent_change, double high_threshold, do
 
 		/* should we send a recovery notification? */
 		if(hst->check_flapping_recovery_notification == TRUE && hst->current_state == HOST_UP) {
-			host_notification(hst, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE);
-			/* Similar to what happens when a host recovers in handle_async_host_check_result */
-			hst->current_notification_number = 0;
-			hst->notified_on = 0;
+			host_notification(hst, NOTIFICATION_RECOVERY, NULL, NULL, NOTIFICATION_OPTION_NONE);
 		}
 	}
 
@@ -663,10 +657,7 @@ void handle_host_flap_detection_disabled(host *hst) {
 
 		/* should we send a recovery notification? */
 		if(hst->check_flapping_recovery_notification == TRUE && hst->current_state == HOST_UP) {
-			host_notification(hst, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE);
-			/* Similar to what happens when a host recovers in handle_async_host_check_result */
-			hst->current_notification_number = 0;
-			hst->notified_on = 0;
+			host_notification(hst, NOTIFICATION_RECOVERY, NULL, NULL, NOTIFICATION_OPTION_NONE);
 		}
 
 		/* clear the recovery notification flag */
@@ -781,10 +772,7 @@ void handle_service_flap_detection_disabled(service *svc) {
 
 		/* should we send a recovery notification? */
 		if(svc->check_flapping_recovery_notification == TRUE && svc->current_state == STATE_OK) {
-			service_notification(svc, NOTIFICATION_NORMAL, NULL, NULL, NOTIFICATION_OPTION_NONE);
-			/* Similar to what happens when a service recovers in handle_async_service_check_result */
-			svc->current_notification_number = 0;
-			svc->notified_on = 0;
+			service_notification(svc, NOTIFICATION_RECOVERY, NULL, NULL, NOTIFICATION_OPTION_NONE);
 		}
 
 		/* clear the recovery notification flag */

base/notifications.c

@@ -98,14 +98,19 @@ int service_notification(service *svc, int type, char *not_author, char *not_dat
 	/* check the viability of sending out a service notification */
 	if(check_service_notification_viability(svc, type, options) == ERROR) {
 		log_debug_info(DEBUGL_NOTIFICATIONS, 0, "Notification viability test failed.  No notification will be sent out.\n");
+		// Recovery notification wasn't sent because they don't want it. Still have to modify values
+		if(type == NOTIFICATION_RECOVERY && ! flag_isset(svc->notification_options, OPT_RECOVERY)) {
+			svc->notified_on = 0;
+			svc->current_notification_number = 0;
+			}
 		/* Set next_notification time if we're in a downtime and
 			notification_interval = zero, otherwise if the service
 			doesn't recover after downtime ends, it will never send
 			the notification */
-			if (svc->notification_interval == 0.0) {
-				if (temp_host->scheduled_downtime_depth > 0 || svc->scheduled_downtime_depth > 0)
-					svc->next_notification = current_time;
-				}
+		if (svc->notification_interval == 0.0) {
+			if (temp_host->scheduled_downtime_depth > 0 || svc->scheduled_downtime_depth > 0)
+				svc->next_notification = current_time;
+			}
 		return ERROR;
 		}
 
@@ -212,7 +217,7 @@ int service_notification(service *svc, int type, char *not_author, char *not_dat
 			mac.x[MACRO_NOTIFICATIONTYPE] = "DOWNTIMECANCELLED";
 		else if(type == NOTIFICATION_CUSTOM)
 			mac.x[MACRO_NOTIFICATIONTYPE] = "CUSTOM";
-		else if(svc->current_state == STATE_OK)
+		else if(type == NOTIFICATION_RECOVERY)
 			mac.x[MACRO_NOTIFICATIONTYPE] = "RECOVERY";
 		else
 			mac.x[MACRO_NOTIFICATIONTYPE] = "PROBLEM";
@@ -276,7 +281,7 @@ int service_notification(service *svc, int type, char *not_author, char *not_dat
 		clear_servicegroup_macros_r(&mac);
 		clear_datetime_macros_r(&mac);
 
-		if(type == NOTIFICATION_NORMAL) {
+		if(type == NOTIFICATION_NORMAL || type == NOTIFICATION_RECOVERY) {
 
 			/* adjust last/next notification time and notification flags if we notified someone */
 			if(contacts_notified > 0) {
@@ -289,8 +294,14 @@ int service_notification(service *svc, int type, char *not_author, char *not_dat
 				/* update the last notification time for this service (this is needed for rescheduling later notifications) */
 				svc->last_notification = current_time;
 
-				/* update notifications flags */
-				add_notified_on(svc, svc->current_state);
+				/* update notifications flags, make sure to consider recovery */
+				if (type == NOTIFICATION_NORMAL) {
+					add_notified_on(svc, svc->current_state);
+					} 
+				else {
+					svc->notified_on = 0;
+					svc->current_notification_number = 0;
+					}
 				}
 
 			/* we didn't end up notifying anyone */
@@ -603,6 +614,20 @@ int check_service_notification_viability(service *svc, int type, int options) {
 		return ERROR;
 		}
 
+	/* If any of the parents are down, don't notify */
+	if (temp_host->parent_hosts != NULL) {
+		hostsmember *temp_hostsmember = NULL;
+		host *parent_host = NULL;
+
+		for(temp_hostsmember = temp_host->parent_hosts; temp_hostsmember != NULL; temp_hostsmember = temp_hostsmember->next) {
+			parent_host = temp_hostsmember->host_ptr;
+			if (parent_host->current_state != HOST_UP) {
+				log_debug_info(DEBUGL_NOTIFICATIONS, 1, "At least one parent (%s) is down, so we won't notify about this service.\n", parent_host->name);
+				return ERROR;
+			}
+		}
+	}
+
 	/* don't notify if we haven't waited long enough since the last time (and the service is not marked as being volatile) */
 	if((current_time < svc->next_notification) && svc->is_volatile == FALSE) {
 		log_debug_info(DEBUGL_NOTIFICATIONS, 1, "We haven't waited long enough to re-notify contacts about this service.\n");
@@ -1065,6 +1090,11 @@ int host_notification(host *hst, int type, char *not_author, char *not_data, int
 
 	/* check viability of sending out a host notification */
 	if(check_host_notification_viability(hst, type, options) == ERROR) {
+		// Recovery notification wasn't sent because they don't want it. Still have to modify values
+		if(type == NOTIFICATION_RECOVERY && ! flag_isset(hst->notification_options, OPT_RECOVERY)) {
+			hst->notified_on = 0;
+			hst->current_notification_number = 0;
+			}
 		log_debug_info(DEBUGL_NOTIFICATIONS, 0, "Notification viability test failed.  No notification will be sent out.\n");
 		return ERROR;
 		}
@@ -1169,7 +1199,7 @@ int host_notification(host *hst, int type, char *not_author, char *not_data, int
 			mac.x[MACRO_NOTIFICATIONTYPE] = "DOWNTIMECANCELLED";
 		else if(type == NOTIFICATION_CUSTOM)
 			mac.x[MACRO_NOTIFICATIONTYPE] = "CUSTOM";
-		else if(hst->current_state == HOST_UP)
+		else if(type == NOTIFICATION_RECOVERY)
 			mac.x[MACRO_NOTIFICATIONTYPE] = "RECOVERY";
 		else
 			mac.x[MACRO_NOTIFICATIONTYPE] = "PROBLEM";
@@ -1233,7 +1263,7 @@ int host_notification(host *hst, int type, char *not_author, char *not_data, int
 		clear_hostgroup_macros_r(&mac);
 		clear_datetime_macros_r(&mac);
 
-		if(type == NOTIFICATION_NORMAL) {
+		if(type == NOTIFICATION_NORMAL || type == NOTIFICATION_RECOVERY) {
 
 			/* adjust last/next notification time and notification flags if we notified someone */
 			if(contacts_notified > 0) {
@@ -1245,7 +1275,13 @@ int host_notification(host *hst, int type, char *not_author, char *not_data, int
 				hst->last_notification = current_time;
 
 				/* update notifications flags */
-				add_notified_on(hst, hst->current_state);
+				if(type == NOTIFICATION_NORMAL) {
+					add_notified_on(hst, hst->current_state);
+					}
+				else {
+					hst->notified_on = 0;
+					hst->current_notification_number = 0;
+					}
 
 				log_debug_info(DEBUGL_NOTIFICATIONS, 0, "%d contacts were notified.  Next possible notification time: %s", contacts_notified, ctime(&hst->next_notification));
 				}
@@ -1449,21 +1485,13 @@ int check_host_notification_viability(host *hst, int type, int options) {
 		}
 
 	/* see if we should notify about problems with this host */
-	if((hst->notification_options & (1 << hst->current_state)) == FALSE) {
+	if(should_notify(hst) == FALSE) {
 		log_debug_info(DEBUGL_NOTIFICATIONS, 1, "We shouldn't notify about %s status for this host.\n", host_state_name(hst->current_state));
 		return ERROR;
 		}
-	if(hst->current_state == HOST_UP) {
-
-		if((hst->notification_options & OPT_RECOVERY) == FALSE) {
-			log_debug_info(DEBUGL_NOTIFICATIONS, 1, "We shouldn't notify about RECOVERY states for this host.\n");
-			return ERROR;
-			}
-		if(!hst->notified_on) {
-			log_debug_info(DEBUGL_NOTIFICATIONS, 1, "We shouldn't notify about this recovery.\n");
-			return ERROR;
-			}
-
+	if(hst->current_state == HOST_UP && hst->notified_on == 0) {
+		log_debug_info(DEBUGL_NOTIFICATIONS, 1, "We shouldn't notify about this recovery.\n");
+		return ERROR;
 		}
 
 	/* see if enough time has elapsed for first notification (Mathias Sundman) */
@@ -1510,6 +1538,20 @@ int check_host_notification_viability(host *hst, int type, int options) {
 		return ERROR;
 		}
 
+	/* If any of the parents are down, don't notify */
+	if (hst->parent_hosts != NULL) {
+		hostsmember *temp_hostsmember = NULL;
+		host *parent_host = NULL;
+
+		for(temp_hostsmember = hst->parent_hosts; temp_hostsmember != NULL; temp_hostsmember = temp_hostsmember->next) {
+			parent_host = temp_hostsmember->host_ptr;
+			if (parent_host->current_state != HOST_UP) {
+				log_debug_info(DEBUGL_NOTIFICATIONS, 1, "At least one parent (%s) is down, so we won't notify about this host.\n", parent_host->name);
+				return ERROR;
+			}
+		}
+	}
+
 	return OK;
 	}
 

include/nagios.h

@@ -326,6 +326,7 @@ extern struct load_control loadctl;
 #define NOTIFICATION_DOWNTIMEEND        6
 #define NOTIFICATION_DOWNTIMECANCELLED  7
 #define NOTIFICATION_CUSTOM             8
+#define NOTIFICATION_RECOVERY           16
 
 
 

t-tap/.gitignore

@@ -8,4 +8,5 @@ test_xsddefault
 test_commands
 test_downtime
 test_strtoul
+test_notifications
 *.dSYM

t-tap/Makefile.in

@@ -24,6 +24,7 @@ TESTS += test_downtime
 TESTS += test_nagios_config
 TESTS += test_timeperiods
 TESTS += test_macros
+TESTS += test_notifications
 
 XSD_OBJS = $(BLD_CGI)/statusdata-cgi.o $(BLD_CGI)/xstatusdata-cgi.o
 XSD_OBJS += $(BLD_CGI)/objects-cgi.o $(BLD_CGI)/xobjects-cgi.o
@@ -104,6 +105,9 @@ test_timeperiods: test_timeperiods.o $(TP_OBJS) $(TAPOBJ)
 test_macros: test_macros.o $(TP_OBJS) $(TAPOBJ)
 	$(CC) $(CFLAGS) -o $@ $^ $(BLD_BASE)/checks.o $(BROKER_LDFLAGS) $(LDFLAGS) $(MATHLIBS) $(SOCKETLIBS) $(LIBS)
 
+test_notifications: test_notifications.o $(BLD_BASE)/notifications.o $(BLD_BASE)/checks.o $(BLD_BASE)/utils.o $(BLD_COMMON)/shared.o $(BLD_BASE)/objects-base.o $(TAPOBJ)
+	$(CC) $(CFLAGS) -o $@ $^ $(LIBS) $(MATHLIBS)
+
 test_xsddefault: test_xsddefault.o $(XSD_OBJS) $(TAPOBJ)
 	$(CC) $(CFLAGS) -o $@ $^ $(LDFLAGS)
 

t-tap/stub_broker.c

@@ -42,3 +42,12 @@ void broker_timed_event(int type, int flags, int attr, timed_event *event, struc
 
 void broker_flapping_data(int type, int flags, int attr, int flapping_type, void *data, double percent_change, double high_threshold, double low_threshold, struct timeval *timestamp)
 { }
+
+int broker_contact_notification_data(int type, int flags, int attr, int notification_type, int reason_type, struct timeval start_time, struct timeval end_time, void *data, contact *cntct, char *ack_author, char *ack_data, int escalated, struct timeval *timestamp)
+{ return OK; }
+
+int broker_contact_notification_method_data(int type, int flags, int attr, int notification_type, int reason_type, struct timeval start_time, struct timeval end_time, void *data, contact *cntct, char *cmd, char *ack_author, char *ack_data, int escalated, struct timeval *timestamp)
+{ return OK; }
+
+int broker_notification_data(int type, int flags, int attr, int notification_type, int reason_type, struct timeval start_time, struct timeval end_time, void *data, char *ack_author, char *ack_data, int escalated, int contacts_notified, struct timeval *timestamp)
+{ return OK; }