Issues with Memory Request value updates for Memory Request Deviation metrices

Products

• DX Application Performance Management

Issue/Introduction

Memory Usage metrics in the DX APM graph: 265207808 bytes ~= 253 MB

Memory Usage in GKE graph around same time: 210.6 MB

There is still 40-50 MB difference which is triggering deviation alarms.  

We upgraded the GKE to 1.28.6 but we still see this issue.

The Kubernetes Pod Memory Usage metrics is reporting incorrect data than being shown in the GKE console due to which we continue to get false alerts.

Also, we can't find the new "Memory Working Set" metric, as suggested earlier.

Resolution

Found the "Memory Working Set" metrics under pods and containers.  The customer confirmed that these metrics appeared to report memory usages more relevant to them.  They would work with their admin to update their alert configurations to use these metrics for reporting alerts on memory usages at the nodes level.