we would like to report a problem with the Azure VM metric:
Virtual Machines - Available Memory Percentage (AVG)
The issue is that when the metric does not have a currently collected/measured value, DX APM reports the value as 0 instead of last measured value.
In our environment the Azure APM agent polling interval is set to 10 minutes. Because of this, the metric shows a real value only at the actual polling time, but the other datapoints between polling intervals are displayed/reported as 0.
This causes false alarm messages.
Please go to this dir <apmia installation dir>/extensions/azure*/config/:
Modify the azure_vm_schema.json, for
name":"Available Memory Percentage (Avg)",
"calculation":"",
"xml_ns":"",
"attributes":{
"apm":{
"apm_type":"LongCounter", -Change this line from LongAverage to LongCounter.
"metric_path":[
{
"name":"Azure"
},
{
"name":"Subscription ID"
},
{
"name":"%azure_subscriptionid"
},
{
"name":"Resource Groups"
},
{
"name":"%resgroup"
},
{
"name":"Virtual Machine"
},
{
"name":"%virtualMachine"
},
{
"name":"Memory"
},
{
"name":"Available Memory Percentage (Avg)"
}
]
}
},
"value":"$.value[?(@.name.value == 'Available Memory Percentage')].['timeseries'].[*].['data'].[0].['average']",
"url":"vmmetric2",
"group":"AZURE VM"
},
Restart APMIA and see if the last value is preserved on less non zero value for that inerval.