[Bug] JMX and Discovery failure with particular JVM flags

[andrewazores]

As reported in #1516, setting the below JVM flags on a target application leads to successful discovery of the application by Cryostat, and successful attempts to open a JMX connection:

-Dcom.sun.management.jmxremote
-Dcom.sun.management.jmxremote.port=9091
-Dcom.sun.management.jmxremote.authenticate=false
-Dcom.sun.management.jmxremote.ssl=false
-Dcom.sun.management.jmxremote.autodiscovery=true

Adding the following additional flags breaks it, so that Cryostat no longer displays the application in the UI:

-Dcom.sun.management.jmxremote.rmi.port=9091
-Djava.rmi.server.hostname=127.0.0.1

[andrewazores]

I believe that adding these two flags is likely not directly breaking Cryostat's discovery of the application - it should still be picking it up by the KubeApi mechanism in the original report's scenario. However, Cryostat will try to open a connection to the JVM to determine its jvmId. Something about this JVM flag configuration causes the JMX connection attempt to fail in a mode that causes Cryostat to reject the discovery node update (as opposed to keeping it with jvmId: null as it does in the case of JMX authentication failure).

[andrewazores]

More information on my tests:

• To connect to the pod using JMC 8 the following 2 args are required:

-Dcom.sun.management.jmxremote.rmi.port=9091
-Djava.rmi.server.hostname=127.0.0.1

• Removing -Djava.rmi.server.hostname=127.0.0.1 makes Cryostat discover the pod, but JMC 8 no longer can connect to the JFR on the App. The bug is triggered when trying to start the recording.

Originally posted by @rjbaucells in #1516 (comment)