Windows Server 2016 WMI performance counters memory leak

[romanblachman]

We have noticed that due to WMI calls used for monitoring as part of telegraf engine, WMI calls to extract performance counters are causing leak in the process. When looking at telegraf project code, I noticed that it is using pdh from this project, and not issuing WMI calls for performance counters directly.

More issues can be found @ influxdata/telegraf#6807

I didn't find direct calls to CoInitializeEx which tells me that PDH is someone calling that internally? Was this problem ever reported to you?