MXS-1262: Use monitor journals in all monitors

All monitors now persist the state of the server in a monitor journal
file.

Moved the removal of stale journals into the core and removed them from
the monitor journal interface.

include/maxscale/monitor.h

@@ -340,20 +340,4 @@ void store_server_journal(MXS_MONITOR *monitor, MXS_MONITOR_SERVERS *master);
  */
 void load_server_journal(MXS_MONITOR *monitor, MXS_MONITOR_SERVERS **master);
 
-/**
- * @brief Remove stored journal file
- *
- * @param monitor Monitor whose journal is removed
- */
-void remove_server_journal(MXS_MONITOR *monitor);
-
-/**
- * @brief Check whether the journal is too old
- *
- * @param monitor Monitor to check
- * @return True if journal is stale or an error occurred while reading the file.
- * False if the file is still valid.
- */
-bool journal_is_stale(MXS_MONITOR *monitor, time_t max_age);
-
 MXS_END_DECLS

server/core/monitor.cc

@@ -75,6 +75,8 @@ static MXS_MONITOR  *allMonitors = NULL;
 static SPINLOCK monLock = SPINLOCK_INIT;
 
 static void monitor_server_free_all(MXS_MONITOR_SERVERS *servers);
+static void remove_server_journal(MXS_MONITOR *monitor);
+static bool journal_is_stale(MXS_MONITOR *monitor, time_t max_age);
 
 /** Server type specific bits */
 static unsigned int server_type_bits = SERVER_MASTER | SERVER_SLAVE |
@@ -190,6 +192,12 @@ monitorStart(MXS_MONITOR *monitor, const MXS_CONFIG_PARAMETER* params)
     {
         spinlock_acquire(&monitor->lock);
 
+        if (journal_is_stale(monitor, monitor->journal_max_age))
+        {
+            MXS_WARNING("Removing stale journal file for monitor '%s'.", monitor->name);
+            remove_server_journal(monitor);
+        }
+
         if ((monitor->handle = (*monitor->module->startMonitor)(monitor, params)))
         {
             monitor->state = MONITOR_STATE_RUNNING;
@@ -1901,6 +1909,8 @@ static const char* process_server(MXS_MONITOR *monitor, const char *data, const
             sptr++;
 
             uint32_t state = sptr[0] | (sptr[1] << 8) | (sptr[2] << 16) | (sptr[3] << 24);
+            db->mon_prev_status = state;
+            db->server->status_pending = state;
             server_set_status_nolock(db->server, state);
             monitor_set_pending_status(db, state);
             break;
@@ -1918,6 +1928,8 @@ static const char* process_server(MXS_MONITOR *monitor, const char *data, const
 static const char* process_master(MXS_MONITOR *monitor, MXS_MONITOR_SERVERS **master, const char *data,
                                   const char *end)
 {
+    if (master)
+    {
         for (MXS_MONITOR_SERVERS* db = monitor->databases; db; db = db->next)
         {
             if (strcmp(db->server->unique_name, data) == 0)
@@ -1926,6 +1938,7 @@ static const char* process_master(MXS_MONITOR *monitor, MXS_MONITOR_SERVERS **ma
                 break;
             }
         }
+    }
 
     data += strlen(data) + 1;
 
@@ -2114,7 +2127,7 @@ void load_server_journal(MXS_MONITOR *monitor, MXS_MONITOR_SERVERS **master)
     }
 }
 
-void remove_server_journal(MXS_MONITOR *monitor)
+static void remove_server_journal(MXS_MONITOR *monitor)
 {
     char path[PATH_MAX];
 
@@ -2128,7 +2141,7 @@ void remove_server_journal(MXS_MONITOR *monitor)
     }
 }
 
-bool journal_is_stale(MXS_MONITOR *monitor, time_t max_age)
+static bool journal_is_stale(MXS_MONITOR *monitor, time_t max_age)
 {
     bool is_stale = true;
     char path[PATH_MAX];

server/modules/monitor/auroramon/auroramon.c

@@ -118,6 +118,8 @@ monitorMain(void *arg)
         return;
     }
 
+    load_server_journal(monitor, NULL);
+
     while (!handle->shutdown)
     {
         lock_monitor_servers(monitor);
@@ -141,6 +143,7 @@ monitorMain(void *arg)
         mon_process_state_changes(monitor, handle->script, handle->events);
 
         servers_status_current_to_pending(monitor);
+        store_server_journal(monitor, NULL);
         release_monitor_servers(monitor);
 
         /** Sleep until the next monitoring interval */

server/modules/monitor/galeramon/galeramon.c

@@ -497,6 +497,7 @@ monitorMain(void *arg)
         return;
     }
     handle->status = MXS_MONITOR_RUNNING;
+    load_server_journal(mon, NULL);
 
     while (1)
     {
@@ -651,6 +652,7 @@ monitorMain(void *arg)
             update_sst_donor_nodes(mon, is_cluster);
         }
 
+        store_server_journal(mon, NULL);
         release_monitor_servers(mon);
     }
 }

server/modules/monitor/mmmon/mmmon.c

@@ -514,6 +514,8 @@ monitorMain(void *arg)
     }
 
     handle->status = MXS_MONITOR_RUNNING;
+    load_server_journal(mon, &handle->master);
+
     while (1)
     {
         if (handle->shutdown)
@@ -616,6 +618,7 @@ monitorMain(void *arg)
 
         mon_hangup_failed_servers(mon);
         servers_status_current_to_pending(mon);
+        store_server_journal(mon, handle->master);
         release_monitor_servers(mon);
     }
 }

server/modules/monitor/mysqlmon.h

@@ -78,7 +78,6 @@ typedef struct
                                    down before failover is initiated */
     bool allow_cluster_recovery; /**< Allow failed servers to rejoin the cluster */
     bool warn_failover; /**< Log a warning when failover happens */
-    bool load_journal; /**< Whether journal file should be loaded */
     bool allow_external_slaves; /**< Whether to allow usage of external slave servers */
     MXS_MONITOR* monitor;
 } MYSQL_MONITOR;

server/modules/monitor/mysqlmon/mysql_mon.c

@@ -244,7 +244,6 @@ startMonitor(MXS_MONITOR *monitor, const MXS_CONFIG_PARAMETER* params)
         handle->shutdown = 0;
         handle->id = config_get_global_options()->id;
         handle->warn_failover = true;
-        handle->load_journal = true;
         handle->monitor = monitor;
     }
 
@@ -263,12 +262,6 @@ startMonitor(MXS_MONITOR *monitor, const MXS_CONFIG_PARAMETER* params)
     handle->events = config_get_enum(params, "events", mxs_monitor_event_enum_values);
     handle->allow_external_slaves = config_get_bool(params, "allow_external_slaves");
 
-    if (journal_is_stale(monitor, monitor->journal_max_age))
-    {
-        MXS_WARNING("Removing stale journal file.");
-        remove_server_journal(monitor);
-    }
-
     bool error = false;
 
     if (!check_monitor_permissions(monitor, "SHOW SLAVE STATUS"))
@@ -1129,7 +1122,9 @@ monitorMain(void *arg)
         MXS_ERROR("mysql_thread_init failed in monitor module. Exiting.");
         return;
     }
+
     handle->status = MXS_MONITOR_RUNNING;
+    load_server_journal(mon, &handle->master);
 
     while (1)
     {
@@ -1169,12 +1164,6 @@ monitorMain(void *arg)
         lock_monitor_servers(mon);
         servers_status_pending_to_current(mon);
 
-        if (handle->load_journal)
-        {
-            handle->load_journal = false;
-            load_server_journal(mon, &handle->master);
-        }
-
         /* start from the first server in the list */
         ptr = mon->databases;
 

server/modules/monitor/ndbclustermon/ndbclustermon.c

@@ -316,7 +316,9 @@ monitorMain(void *arg)
         MXS_ERROR("Fatal : mysql_thread_init failed in monitor module. Exiting.");
         return;
     }
+
     handle->status = MXS_MONITOR_RUNNING;
+    load_server_journal(mon, NULL);
 
     while (1)
     {
@@ -374,6 +376,7 @@ monitorMain(void *arg)
 
         mon_hangup_failed_servers(mon);
         servers_status_current_to_pending(mon);
+        store_server_journal(mon, NULL);
         release_monitor_servers(mon);
     }
 }