Merge branch 'feature/ido-ha-6827'

fixes #6827
refs #6203

components/db_ido_mysql/idomysqlconnection.cpp

@@ -33,7 +33,7 @@
 
 using namespace icinga;
 
-#define SCHEMA_VERSION "1.11.6"
+#define SCHEMA_VERSION "1.11.7"
 
 REGISTER_TYPE(IdoMysqlConnection);
 REGISTER_STATSFUNCTION(IdoMysqlConnectionStats, &IdoMysqlConnection::StatsFunc);
@@ -188,6 +188,7 @@ void IdoMysqlConnection::Reconnect(void)
 		passwd = (!ipasswd.IsEmpty()) ? ipasswd.CStr() : NULL;
 		db = (!idb.IsEmpty()) ? idb.CStr() : NULL;
 
+		/* connection */
 		if (!mysql_init(&m_Connection)) {
 			std::ostringstream msgbuf;
 			msgbuf << "mysql_init() failed: \"" << mysql_error(&m_Connection) << "\"";
@@ -210,9 +211,9 @@ void IdoMysqlConnection::Reconnect(void)
 		String dbVersionName = "idoutils";
 		IdoMysqlResult result = Query("SELECT version FROM " + GetTablePrefix() + "dbversion WHERE name='" + Escape(dbVersionName) + "'");
 
-		Dictionary::Ptr version_row = FetchRow(result);
+		Dictionary::Ptr row = FetchRow(result);
 
-		if (!version_row) {
+		if (!row) {
 			Log(LogCritical, "IdoMysqlConnection", "Schema does not provide any valid version! Verify your schema installation.");
 
 			Application::Exit(EXIT_FAILURE);
@@ -220,7 +221,7 @@ void IdoMysqlConnection::Reconnect(void)
 
 		DiscardRows(result);
 
-		String version = version_row->Get("version");
+		String version = row->Get("version");
 
 		if (Utility::CompareVersion(SCHEMA_VERSION, version) < 0) {
 			Log(LogCritical, "IdoMysqlConnection", "Schema version '" + version + "' does not match the required version '" +
@@ -232,16 +233,68 @@ void IdoMysqlConnection::Reconnect(void)
 		String instanceName = GetInstanceName();
 
 		result = Query("SELECT instance_id FROM " + GetTablePrefix() + "instances WHERE instance_name = '" + Escape(instanceName) + "'");
-
-		Dictionary::Ptr row = FetchRow(result);
+		row = FetchRow(result);
 
 		if (!row) {
 			Query("INSERT INTO " + GetTablePrefix() + "instances (instance_name, instance_description) VALUES ('" + Escape(instanceName) + "', '" + Escape(GetInstanceDescription()) + "')");
 			m_InstanceID = GetLastInsertID();
 		} else {
+			m_InstanceID = DbReference(row->Get("instance_id"));
+		}
+
+		DiscardRows(result);
+
+		Endpoint::Ptr my_endpoint = Endpoint::GetLocalEndpoint();
+
+		/* we have an endpoint in a cluster setup, so decide if we can proceed here */
+		if (my_endpoint && GetHAMode() == HARunOnce) {
+			/* get the current endpoint writing to programstatus table */
+			result = Query("SELECT UNIX_TIMESTAMP(status_update_time) AS status_update_time, endpoint_name FROM " +
+			    GetTablePrefix() + "programstatus WHERE instance_id = " + Convert::ToString(m_InstanceID));
+			row = FetchRow(result);
 			DiscardRows(result);
 
-			m_InstanceID = DbReference(row->Get("instance_id"));
+			String endpoint_name;
+
+			if (row)
+				endpoint_name = row->Get("endpoint_name");
+			else
+				Log(LogNotice, "IdoMysqlConnection", "Empty program status table");
+
+			/* if we did not write into the database earlier, another instance is active */
+			if (endpoint_name != my_endpoint->GetName()) {
+				double status_update_time;
+
+				if (row)
+					status_update_time = row->Get("status_update_time");
+				else
+					status_update_time = 0;
+
+				double status_update_age = Utility::GetTime() - status_update_time;
+
+				Log(LogNotice, "IdoMysqlConnection", "Last update by '" +
+				    endpoint_name + "' was " + Convert::ToString(status_update_age) + "s ago.");
+
+				if (status_update_age < GetFailoverTimeout()) {
+					mysql_close(&m_Connection);
+					m_Connected = false;
+
+					return;
+				}
+
+				/* activate the IDO only, if we're authoritative in this zone */
+				if (IsPaused()) {
+					Log(LogNotice, "IdoMysqlConnection", "Local endpoint '" +
+					    my_endpoint->GetName() + "' is not authoritative, bailing out.");
+
+					mysql_close(&m_Connection);
+					m_Connected = false;
+
+					return;
+				}
+			}
+
+			Log(LogNotice, "IdoMysqlConnection", "Enabling IDO connection.");
 		}
 
 		std::ostringstream msgbuf;

components/db_ido_mysql/schema/mysql.sql

@@ -900,6 +900,7 @@ CREATE TABLE IF NOT EXISTS icinga_programstatus (
   status_update_time timestamp  default '0000-00-00 00:00:00',
   program_start_time timestamp  default '0000-00-00 00:00:00',
   program_end_time timestamp  default '0000-00-00 00:00:00',
+  endpoint_name varchar(255) character set latin1 collate latin1_general_cs default NULL,
   is_currently_running smallint default 0,
   process_id bigint unsigned default 0,
   daemon_mode smallint default 0,

components/db_ido_mysql/schema/upgrade/2.1.0.sql

@@ -0,0 +1,17 @@
+-- -----------------------------------------
+-- upgrade path for Icinga 2.1.0
+--
+-- -----------------------------------------
+-- Copyright (c) 2014 Icinga Development Team (http://www.icinga.org)
+--
+-- Please check http://docs.icinga.org for upgrading information!
+-- -----------------------------------------
+
+ALTER TABLE `icinga_programstatus` ADD COLUMN `endpoint_name` varchar(255) character set latin1 collate latin1_general_cs default NULL;
+
+-- -----------------------------------------
+-- update dbversion
+-- -----------------------------------------
+
+INSERT INTO icinga_dbversion (name, version, create_time, modify_time) VALUES ('idoutils', '1.11.7', NOW(), NOW()) ON DUPLICATE KEY UPDATE version='1.11.7', modify_time=NOW();
+

components/db_ido_pgsql/idopgsqlconnection.cpp

@@ -34,7 +34,7 @@
 
 using namespace icinga;
 
-#define SCHEMA_VERSION "1.11.6"
+#define SCHEMA_VERSION "1.11.7"
 
 REGISTER_TYPE(IdoPgsqlConnection);
 
@@ -212,15 +212,15 @@ void IdoPgsqlConnection::Reconnect(void)
 		String dbVersionName = "idoutils";
 		IdoPgsqlResult result = Query("SELECT version FROM " + GetTablePrefix() + "dbversion WHERE name=E'" + Escape(dbVersionName) + "'");
 
-		Dictionary::Ptr version_row = FetchRow(result, 0);
+		Dictionary::Ptr row = FetchRow(result, 0);
 
-		if (!version_row) {
+		if (!row) {
 			Log(LogCritical, "IdoPgsqlConnection", "Schema does not provide any valid version! Verify your schema installation.");
 
 			Application::Exit(EXIT_FAILURE);
 		}
 
-		String version = version_row->Get("version");
+		String version = row->Get("version");
 
 		if (Utility::CompareVersion(SCHEMA_VERSION, version) < 0) {
 			Log(LogCritical, "IdoPgsqlConnection", "Schema version '" + version + "' does not match the required version '" +
@@ -232,8 +232,7 @@ void IdoPgsqlConnection::Reconnect(void)
 		String instanceName = GetInstanceName();
 
 		result = Query("SELECT instance_id FROM " + GetTablePrefix() + "instances WHERE instance_name = E'" + Escape(instanceName) + "'");
-
-		Dictionary::Ptr row = FetchRow(result, 0);
+		row = FetchRow(result, 0);
 
 		if (!row) {
 			Query("INSERT INTO " + GetTablePrefix() + "instances (instance_name, instance_description) VALUES (E'" + Escape(instanceName) + "', E'" + Escape(GetInstanceDescription()) + "')");
@@ -242,6 +241,58 @@ void IdoPgsqlConnection::Reconnect(void)
 			m_InstanceID = DbReference(row->Get("instance_id"));
 		}
 
+		Endpoint::Ptr my_endpoint = Endpoint::GetLocalEndpoint();
+
+		/* we have an endpoint in a cluster setup, so decide if we can proceed here */
+		if (my_endpoint && GetHAMode() == HARunOnce) {
+			/* get the current endpoint writing to programstatus table */
+			result = Query("SELECT UNIX_TIMESTAMP(status_update_time) AS status_update_time, endpoint_name FROM " +
+			    GetTablePrefix() + "programstatus WHERE instance_id = " + Convert::ToString(m_InstanceID));
+			row = FetchRow(result, 0);
+
+			String endpoint_name;
+
+			if (row)
+				endpoint_name = row->Get("endpoint_name");
+			else
+				Log(LogNotice, "IdoPgsqlConnection", "Empty program status table");
+
+			/* if we did not write into the database earlier, another instance is active */
+			if (endpoint_name != my_endpoint->GetName()) {
+				double status_update_time;
+
+				if (row)
+					status_update_time = row->Get("status_update_time");
+				else
+					status_update_time = 0;
+
+				double status_update_age = Utility::GetTime() - status_update_time;
+
+				Log(LogNotice, "IdoPgsqlConnection", "Last update by '" +
+				    endpoint_name + "' was " + Convert::ToString(status_update_age) + "s ago.");
+
+				if (status_update_age < GetFailoverTimeout()) {
+					PQfinish(m_Connection);
+					m_Connection = NULL;
+
+					return;
+				}
+
+				/* activate the IDO only, if we're authoritative in this zone */
+				if (IsPaused()) {
+					Log(LogNotice, "IdoPgsqlConnection", "Local endpoint '" +
+					    my_endpoint->GetName() + "' is not authoritative, bailing out.");
+
+					PQfinish(m_Connection);
+					m_Connection = NULL;
+
+					return;
+				}
+			}
+
+			Log(LogNotice, "IdoPgsqlConnection", "Enabling IDO connection.");
+		}
+
 		std::ostringstream msgbuf;
 		msgbuf << "pgSQL IDO instance id: " << static_cast<long>(m_InstanceID) << " (schema version: '" + version + "')";
 		Log(LogInformation, "IdoPgsqlConnection", msgbuf.str());

components/db_ido_pgsql/schema/pgsql.sql

@@ -928,6 +928,7 @@ CREATE TABLE  icinga_programstatus (
   program_start_time timestamp with time zone default '1970-01-01 00:00:00',
   program_end_time timestamp with time zone default '1970-01-01 00:00:00',
   is_currently_running INTEGER  default 0,
+  endpoint_name TEXT  default '',
   process_id bigint default 0,
   daemon_mode INTEGER  default 0,
   last_command_check timestamp with time zone default '1970-01-01 00:00:00',

components/db_ido_pgsql/schema/upgrade/2.1.0.sql

@@ -0,0 +1,17 @@
+-- -----------------------------------------
+-- upgrade path for Icinga 2.1.0
+--
+-- -----------------------------------------
+-- Copyright (c) 2014 Icinga Development Team (http://www.icinga.org)
+--
+-- Please check http://docs.icinga.org for upgrading information!
+-- -----------------------------------------
+
+ALTER TABLE icinga_programstatus ADD COLUMN endpoint_name TEXT default NULL;
+
+-- -----------------------------------------
+-- update dbversion
+-- -----------------------------------------
+
+SELECT updatedbversion('1.11.7');
+

doc/3-monitoring-basics.md

@@ -1784,6 +1784,8 @@ chapter. Details on the configuration can be found in the
 [IdoMysqlConnection](#objecttype-idomysqlconnection) and
 [IdoPgsqlConnection](#objecttype-idoPgsqlconnection)
 object configuration documentation.
+The DB IDO feature supports [High Availability](##high-availability-db-ido) in
+the Icinga 2 cluster.
 
 The following example query checks the health of the current Icinga 2 instance
 writing its current status to the DB IDO backend table `icinga_programstatus`

doc/4-monitoring-remote-systems.md

@@ -479,6 +479,46 @@ the Icinga 2 daemon.
 
     # icinga2 -c /etc/icinga2/node1/icinga2.conf -DLocalStateDir=/opt/node1/var
 
+### <a id="high-availability-db-ido"></a> High Availability with DB IDO
+
+All instances within the same zone (e.g. the `master` zone as HA cluster) must
+have the DB IDO feature enabled.
+
+Example DB IDO MySQL:
+
+    # icinga2-enable-feature ido-mysql
+    The feature 'ido-mysql' is already enabled.
+
+By default the DB IDO feature only runs on the elected zone master. All other nodes
+disable the active IDO database connection at runtime.
+
+> **Note**
+>
+> The DB IDO HA feature can be disabled by setting the `enable_ha` attribute to `false`
+> for the [IdoMysqlConnection](#objecttype-idomysqlconnection) or
+> [IdoPgsqlConnection](#objecttype-idopgsqlconnection) object on all nodes in the
+> same zone.
+>
+> All endpoints will enable the DB IDO feature then, connect to the configured
+> database and dump configuration, status and historical data on their own.
+
+If the instance with the active DB IDO connection dies, the HA functionality will
+re-enable the DB IDO connection on the newly elected zone master.
+
+The DB IDO feature will try to determine which cluster endpoint is currently writing
+to the database and bail out if another endpoint is active. You can manually verify that
+by running the following query:
+
+    icinga=> SELECT status_update_time, endpoint_name FROM icinga_programstatus;
+       status_update_time   | endpoint_name
+    ------------------------+---------------
+     2014-08-15 15:52:26+02 | icinga2a
+    (1 Zeile)
+
+This is useful when the cluster connection between endpoints breaks, and prevents
+data duplication in split-brain-scenarios. The failover timeout can be set for the
+`failover_timeout` attribute, but not lower than 60 seconds.
+
 
 ### <a id="cluster-scenarios"></a> Cluster Scenarios
 
@@ -630,7 +670,8 @@ High availability with Icinga 2 is possible by putting multiple nodes into
 a dedicated `Zone`. All nodes will elect their active master, and retry an
 election once the current active master failed.
 
-Selected features (such as DB IDO) will only be active on the current active master.
+Selected features (such as [DB IDO](#high-availability-db-ido)) will only be
+active on the current active master.
 All other passive nodes will pause the features without reload/restart.
 
 Connections from other zones will be accepted by all active and passive nodes

doc/6-configuring-icinga-2.md

@@ -1272,6 +1272,8 @@ Attributes:
   table\_prefix   |**Optional.** MySQL database table prefix. Defaults to "icinga\_".
   instance\_name  |**Optional.** Unique identifier for the local Icinga 2 instance. Defaults to "default".
   instance\_description|**Optional.** Description for the Icinga 2 instance.
+  enable_ha       |**Optional.** Enable the high availability functionality. Only valid in a [cluster setup](#high-availability-db-ido). Defaults to "true".
+  failover_timeout | **Optional.** Set the failover timeout in a [HA cluster](#high-availability-db-ido). Must not be lower than 60s". Defaults to "60s".
   cleanup         |**Optional.** Dictionary with items for historical table cleanup.
   categories      |**Optional.** The types of information that should be written to the database.
 
@@ -1359,6 +1361,8 @@ Attributes:
   table\_prefix   |**Optional.** PostgreSQL database table prefix. Defaults to "icinga\_".
   instance\_name  |**Optional.** Unique identifier for the local Icinga 2 instance. Defaults to "default".
   instance\_description|**Optional.** Description for the Icinga 2 instance.
+  enable_ha       |**Optional.** Enable the high availability functionality. Only valid in a [cluster setup](#high-availability-db-ido). Defaults to "true".
+  failover_timeout | **Optional.** Set the failover timeout in a [HA cluster](#high-availability-db-ido). Must not be lower than 60s". Defaults to "60s".
   cleanup         |**Optional.** Dictionary with items for historical table cleanup.
   categories      |**Optional.** The types of information that should be written to the database.
 

lib/base/dynamicobject.ti

@@ -2,6 +2,12 @@ namespace icinga
 {
 
 code {{{
+enum HAMode
+{
+	HARunOnce,
+	HARunEverywhere
+};
+
 class NameComposer {
 public:
 	virtual String MakeName(const String& shortName, const Dictionary::Ptr props) const = 0;
@@ -32,6 +38,7 @@ abstract class DynamicObject
 	[get_protected] bool stop_called;
 	[get_protected] bool pause_called;
 	[get_protected] bool resume_called;
+	[enum] HAMode ha_mode (HAMode);
 	Dictionary::Ptr authority_info;
 	[protected] Dictionary::Ptr extensions;
 

lib/db_ido/db_ido-type.conf

@@ -18,6 +18,8 @@
  ******************************************************************************/
 
 %type DbConnection {
+        %validator "ValidateFailoverTimeout"
+
 	%attribute %string "table_prefix",
 
 	%attribute %dictionary "cleanup" {
@@ -38,5 +40,9 @@
                 %attribute %number "systemcommands_age",
 	},
 
-	%attribute %number "categories"
+	%attribute %number "categories",
+
+        %attribute %number "enable_ha",
+
+        %attribute %number "failover_timeout",
 }

lib/db_ido/dbconnection.cpp

@@ -22,22 +22,35 @@
 #include "icinga/icingaapplication.hpp"
 #include "icinga/host.hpp"
 #include "icinga/service.hpp"
+#include "config/configcompilercontext.hpp"
 #include "base/dynamictype.hpp"
 #include "base/convert.hpp"
 #include "base/objectlock.hpp"
 #include "base/utility.hpp"
 #include "base/initialize.hpp"
 #include "base/logger_fwd.hpp"
+#include "base/scriptfunction.hpp"
 #include <boost/foreach.hpp>
 
 using namespace icinga;
 
 REGISTER_TYPE(DbConnection);
+REGISTER_SCRIPTFUNCTION(ValidateFailoverTimeout, &DbConnection::ValidateFailoverTimeout);
 
 Timer::Ptr DbConnection::m_ProgramStatusTimer;
 
 INITIALIZE_ONCE(&DbConnection::StaticInitialize);
 
+void DbConnection::OnConfigLoaded(void)
+{
+	DynamicObject::OnConfigLoaded();
+
+	if (!GetEnableHa()) {
+		Log(LogDebug, "DbConnection", "HA functionality disabled. Won't pause IDO connection: " + GetName());
+		SetHAMode(HARunEverywhere);
+	}
+}
+
 void DbConnection::Start(void)
 {
 	DynamicObject::Start();
@@ -108,6 +121,7 @@ void DbConnection::ProgramStatusHandler(void)
 	query2.Fields->Set("status_update_time", DbValue::FromTimestamp(Utility::GetTime()));
 	query2.Fields->Set("program_start_time", DbValue::FromTimestamp(Application::GetStartTime()));
 	query2.Fields->Set("is_currently_running", 1);
+	query2.Fields->Set("endpoint_name", IcingaApplication::GetInstance()->GetNodeName());
 	query2.Fields->Set("process_id", Utility::GetPid());
 	query2.Fields->Set("daemon_mode", 1);
 	query2.Fields->Set("last_command_check", DbValue::FromTimestamp(Utility::GetTime()));
@@ -358,7 +372,7 @@ void DbConnection::UpdateAllObjects(void)
 
 void DbConnection::PrepareDatabase(void)
 {
-	/* 
+	/*
 	 * only clear tables on reconnect which
 	 * cannot be updated by their existing ids
 	 * for details check https://dev.icinga.org/issues/5565
@@ -384,7 +398,6 @@ void DbConnection::PrepareDatabase(void)
 	//ClearConfigTable("hostgroups");
 	//ClearConfigTable("hosts");
 	//ClearConfigTable("hoststatus");
-	ClearConfigTable("programstatus");
 	ClearConfigTable("scheduleddowntime");
 	ClearConfigTable("service_contactgroups");
 	ClearConfigTable("service_contacts");
@@ -400,3 +413,12 @@ void DbConnection::PrepareDatabase(void)
 		FillIDCache(type);
 	}
 }
+
+void DbConnection::ValidateFailoverTimeout(const String& location, const Dictionary::Ptr& attrs)
+{
+	Value failover_timeout = attrs->Get("failover_timeout");
+	if (failover_timeout < 60) {
+                ConfigCompilerContext::GetInstance()->AddMessage(true, "Validation failed for " +
+                    location + ": Failover timeout minimum is 60s.");
+	}
+}

lib/db_ido/dbconnection.hpp

@@ -63,7 +63,10 @@ public:
 	void SetStatusUpdate(const DbObject::Ptr& dbobj, bool hasupdate);
 	bool GetStatusUpdate(const DbObject::Ptr& dbobj) const;
 
+        static void ValidateFailoverTimeout(const String& location, const Dictionary::Ptr& attrs);
+
 protected:
+        virtual void OnConfigLoaded(void);
 	virtual void Start(void);
 	virtual void Resume(void);
 	virtual void Pause(void);

lib/db_ido/dbconnection.ti

@@ -21,6 +21,14 @@ abstract class DbConnection : DynamicObject
 			    DbCatLog | DbCatNotification | DbCatProgramStatus | DbCatRetention | DbCatStateHistory;
 		}}}
 	};
+
+	[config] bool enable_ha {
+		default {{{ return true; }}}
+	};
+
+	[config] double failover_timeout {
+		default {{{ return 60; }}}
+	};
 };
 
 }

lib/remote/authority.cpp

@@ -61,7 +61,8 @@ static void AuthorityTimerHandler(void)
 		BOOST_FOREACH(const DynamicObject::Ptr& object, type->GetObjects()) {
 			Endpoint::Ptr endpoint = endpoints[Utility::SDBM(object->GetName()) % endpoints.size()];
 
-			object->SetAuthority(endpoint == my_endpoint);
+			if (object->GetHAMode() == HARunOnce)
+				object->SetAuthority(endpoint == my_endpoint);
 		}
 	}
 }