16867 FIX azure: Remove unnecessary 'metric not found' errors

When querying metrics, Azure agent was reporting each missing metric in the
details of the 'Azure Agent Info' service. This didn't influence the status
of the service but still implied an error.

It was misleading because some metrics aren't defined in some types
of resources. Now, those metrics are optional and won't cause an error.

SUP-19726

Change-Id: If8f370610c32252f0f9499c4bfed4a60a4910203

.werks/16867.md

@@ -0,0 +1,19 @@
+[//]: # (werk v2)
+# azure: Remove unnecessary 'metric not found' errors
+
+key        | value
+---------- | ---
+compatible | yes
+version    | 2.3.0p13
+date       | 2024-08-08T15:40:36+00:00
+level      | 1
+class      | fix
+component  | checks
+edition    | cre
+
+When querying metrics, Azure agent was reporting each missing metric in the
+details of the 'Azure Agent Info' service. This didn't influence the status
+of the service but still implied an error.
+
+It was misleading because some metrics aren't defined in some types
+of resources. Now, those metrics are optional and won't cause an error.

cmk/special_agents/agent_azure.py

@@ -197,6 +197,22 @@
     ],
 }
 
+OPTIONAL_METRICS: Mapping[str, Sequence[str]] = {
+    "Microsoft.Sql/servers/databases": [
+        "storage_percent",
+        "deadlock",
+        "dtu_consumption_percent",
+    ],
+    "Microsoft.DBforMySQL/servers": ["seconds_behind_master"],
+    "Microsoft.DBforMySQL/flexibleServers": ["replication_lag"],
+    "Microsoft.DBforPostgreSQL/servers": ["pg_replica_log_delay_in_seconds"],
+    "Microsoft.Network/loadBalancers": ["AllocatedSnatPorts", "UsedSnatPorts"],
+    "Microsoft.Compute/virtualMachines": [
+        "CPU Credits Consumed",
+        "CPU Credits Remaining",
+    ],
+}
+
 
 class TagsImportPatternOption(enum.Enum):
     ignore_all = "IGNORE_ALL"
@@ -1340,7 +1356,7 @@ def get_validity_from_args(self, *args: Any) -> bool:
         return True
 
     def get_live_data(self, *args: Any) -> Any:
-        mgmt_client, resource_id, err = args
+        mgmt_client, resource_id, resource_type, err = args
         metricnames, interval, aggregation, filter_ = self.metric_definition
 
         raw_metrics = mgmt_client.metrics(
@@ -1358,7 +1374,11 @@ def get_live_data(self, *args: Any) -> Any:
             if parsed_metric is not None:
                 metrics.append(parsed_metric)
             else:
-                msg = "metric not found: {} ({})".format(raw_metric["name"]["value"], aggregation)
+                metric_name = raw_metric["name"]["value"]
+                if metric_name in OPTIONAL_METRICS.get(resource_type, []):
+                    continue
+
+                msg = "metric not found: {} ({})".format(metric_name, aggregation)
                 err.add("info", resource_id, msg)
                 LOGGER.info(msg)
 
@@ -1408,7 +1428,11 @@ def gather_metrics(mgmt_client, resource, debug=False):
         cache = MetricCache(resource, metric_def, NOW, debug=debug)
         try:
             resource.metrics += cache.get_data(
-                mgmt_client, resource.info["id"], err, use_cache=cache.cache_interval > 60
+                mgmt_client,
+                resource.info["id"],
+                resource.info["type"],
+                err,
+                use_cache=cache.cache_interval > 60,
             )
         except ApiError as exc:
             if debug: