Cryostat Observations and Questions: JVM Target Detection, Custom Metrics, and Recording Visualization

[Shreya Biradar]

Hi team, I tried collecting some profiles and JFR recordings for a Quarkus application using Cryostat, and here are a few observations and questions:

1. At times Cryostat fails to automatically detect the JVM target when the pod restarts, is this a bug or expected behavior due to some connection issue?
   
   
2. How can custom metrics/event data be added for visualization in the integrated Grafana dashboard for example - by default  'jdk.GCHeapSummary.heapUsed' is used to display Heap usage data, similarly which metric should be used to visualize GC Pause time, and is there a way to extract JVM metrics available, which can be visualized using the inbuilt Grafana dashboard by Cryostat.
   
   
3. In the case of multiple JFR recordings, when I try to view a particular recording using the  integrated Grafana dashboard except for the first recording, the consecutive recordings display the recording start time as that of the first recording.
   For example, I have 2 recordings-
   Recording1 - Start time: 2023-06-28 16:08:47 and Duration: 30s
   Recording2 - Start time: 2023-06-28 16:15:53 and Duration: 63s
   When I try to visualize Recording2 using the integrated Grafana dashboard, Recording Start time: 2023-06-28 16:08:47 and Recording Duration: 8.2 min is displayed, which contains data from Recording1 as well. Why is it so and is there a way to visualize a particular recording based on its start time?

Thank you.
Shreya