Average aggregated 24 hours usage of Windows CPU data collected by Metricbeat exceeds some Thousand but after restart of Metricbeat gives accurate aggregated CPU data.
We have 100s of Windows Server, every week experiencing the same issue randomly for 4 to 5 servers on average.
Aggregated average data (24 hrs of data) of CPU utilization is abnormal like 20945.584% but after restarting Metricbeat issue gets resolved.
Everytime We are restarting Metricbeat in affected Windows Servers.
For affected servers, 15 mins or 30 mins data are accurate, only aggregated data result is abnormal.
What would be the root cause? How to Mitigate this issue?