Metricbeat reporting incorrect system.diskio.iostat.queue.avg_size values

I'm using Elastic's machine learning to detect anomalies in system usage data, one of the fields being system.diskio.iostat.queue.avg_size. The data we are receiving occasionally reports anomalies in the disk IO average queue size that are incorrect and not possible.

As an example a particular server reports a typical value of 0.902 for this field but the actual value for the anomaly was 2,285,852,514,112.815.

This same anomaly has occurred across other systems, another one reporting a typical value of 0.684 and actual value of 1,091,778,685,151.05.

Is this an issue in how metricbeat is collecting system.diskio.iostat.queue.avg_size data? How can this issue be fixed?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.