App Crash Reported in Events but Not in crashingLRPs Metric

Products

• VMware Tanzu Application Service

Issue/Introduction

 

• Running `cf events <app-name>` shows multiple app.crash events.

• The HealthWatch dashboard or cfdot reports crashingLRPs as 0.

• The application may appear to be in a starting or running state despite recent crashes.

 

 

Resolution

The discrepancy is caused by how the two layers of Elastic Application Runtime track failures:

• Diego crashingLRPs (Real-time State): This is a point-in-time snapshot. Because Diego is self-healing, it transitions a LRP from CRASHED to UNCLAIMED, then CLAIMED almost instantly. Unless your monitoring tool scrapes the metric at the exact second of the failure, it will report 0.

• Cloud Controller Events (Persistent History): Every crash triggers a permanent log entry (app.crash). These records remain in the database even after the app has recovered or moved into a restart delay, creating a historical trail that the real-time metric doesn't show.