Availability check was greatly improved by changes introduced in https://issues.jboss.org/browse/ISPN-4236 and https://issues.jboss.org/browse/ISPN-4235.

This works pretty well in case of one deployment of an embedded application on one application server, which is monitored by an agent.

However, when we try to monitor more deployments by the same agent (app servers, switched ports, each one with its own deployment), the process of availability checking is unstable.

Cache managers / caches... are reported repeatedly UP and DOWN, UP and DOWN. See linked bugzilla with attached screenshots depicting this issue.

I suspect higher number of MBeans to cause this problem.

When we deploy on 2 servers, one of them for example was able to stabilize availability check after 2 hours and then, it stays UP.

During deployment in 2xdomain mode = 4 servers, availability check "is skipping, jumping" UP, DOWN, UP, DOWN... etc. again, see screenshots in BZ.