Merge pull request #991 from thomassa/pr1221

CP-2190 for PR-1221: fix memory monitoring

Author: Jon Ludlam

ocaml/rrdd/rrdd_main.ml

@@ -427,7 +427,7 @@ let read_mem_metrics xc =
 	let xapiactualfree_kib = Int64.of_int (gcstat.Gc.free_words / 256) in
 	let xapiactuallive_kib = Int64.of_int (gcstat.Gc.live_words / 256) in
 	[
-		(Host, ds_make ~name:"memory_total_kib" ~description:"Total amount of memory in use"
+		(Host, ds_make ~name:"memory_total_kib" ~description:"Total amount of memory in the host"
 		~value:(Rrd.VT_Int64 total_kib) ~ty:Rrd.Gauge ~min:0.0 ~default:true ~units:"KiB" ());
 		(Host, ds_make ~name:"memory_free_kib" ~description:"Total amount of free memory"
 		 ~value:(Rrd.VT_Int64 free_kib) ~ty:Rrd.Gauge ~min:0.0 ~default:true ~units:"KiB" ());

scripts/perfmon

@@ -404,7 +404,7 @@ class VariableConfig:
         try:
             self.alarm_trigger_period = int(alarm_trigger_period)
         except:
-            raise XmlConfigException, "variable %s: alarm_trigger_level %s not an int" % \
+            raise XmlConfigException, "variable %s: alarm_trigger_period %s not an int" % \
                 (name, alarm_trigger_period)
 
         try:
@@ -655,7 +655,7 @@ class HOSTMonitor(ObjectMonitor):
        * consolidation_fn: how to combine variables from rrd_updates into one value
          (default is 'average' for 'cpu_usage' & 'sum' for everything else)
        * rrd_regex matches the names of variables from (xe host-data-source-list uuid=$hostuuid) used to compute value
-         (only has defaults for "cpu_usage", "network_usage", "memory_total_kib" and "sr_io_throughput_total_xxxxxxxx")
+         (only has defaults for "cpu_usage", "network_usage", "memory_free_kib" and "sr_io_throughput_total_xxxxxxxx")
     """
     def __init__(self, *args):
         self.monitortype = "Host"
@@ -670,12 +670,14 @@ class HOSTMonitor(ObjectMonitor):
         elif config_tag == 'rrd_regex':
             if variable_name == "cpu_usage": return "cpu[0-9]+"
             elif variable_name == "network_usage": return "pif_eth[0-9]+_[rt]x"
-            elif variable_name == "memory_total_kib": return variable_name
+            elif variable_name == "memory_free_kib": return variable_name
             elif re.match("sr_io_throughput_total_[0-9a-f]{8}$", variable_name): return variable_name[3:]
             else: raise XmlConfigException, "variable %s: no default rrd_regex - please specify one" % variable_name
         elif config_tag == 'alarm_trigger_period':      return '60'   # 1 minute
         elif config_tag == 'alarm_auto_inhibit_period': return '3600' # 1 hour
-        elif config_tag == 'alarm_trigger_sense':       return 'high' # trigger if *above*
+        elif config_tag == 'alarm_trigger_sense':
+            if variable_name == "memory_free_kib": return "low"
+            else: return 'high' # trigger if *above* level
         elif config_tag == 'alarm_priority':             return '5'    # the minimum priority required for mail-alarm to send
         else: raise XmlConfigException, "variable %s: no default available for tag %s" % (variable_name, config_tag)