hpe: Remove redundant calls to enclosurechassis endpoint

On mono-node HPE chassis, the current code is calling the
enclosurechassis endpoint for each monitoring loop.

This call is perfectly useless since the enclosurechassis does not
exists.

This patch is about adding a is_multinode_chassis() function to be
cached that try this endpoint and get a constant answer.

Regarding this cached status, a real read of the enclosurechassis is
done for multinode chassis only.

Tested on DL380, ilorest 4.9.0.0, Model: 'iLO 5' FW: 'iLO 5 v3.03'
Tested on XL225n (Apollo 2000 Gen10), iloreset 4.9.0.0, Model: 'iLO 5' FW: 'iLO 5 v3.08'
Tested on DL340Gen12, ilorest 4.9.0.0, Model: 'iLO 6' FW: 'iLO 6 v1.62'

Signed-off-by: Erwan Velu <[email protected]>

hwbench/environment/vendors/hpe/hpe.py

@@ -168,11 +168,23 @@ def read_power_consumption(
             )
         return power_consumption
 
-    def get_oem_chassis(self):
-        return self.get_redfish_url(
-            "/redfish/v1/Chassis/enclosurechassis/", log_failure=False
+    @cachetools.func.ttl_cache(maxsize=1, ttl=3600)
+    def is_multinode_chassis(self) -> bool:
+        return (
+            True
+            if self.get_redfish_url(
+                "/redfish/v1/Chassis/enclosurechassis/", log_failure=False
+            )
+            else False
         )
 
+    def get_oem_chassis(self):
+        if self.is_multinode_chassis():
+            return self.get_redfish_url(
+                "/redfish/v1/Chassis/enclosurechassis/", log_failure=False
+            )
+        return {}
+
 
 class Hpe(Vendor):
     def __init__(self, out_dir, dmi, monitoring_config_filename):