Current implementation of ClusterHealth.getHealthStatus() only checks the user caches. Internal caches created should have be part of the cluster health and we can check them via InternalCacheRegistry.getInternalCaches().

Also, it should use CacheManager.getCache(cacheName, false) to avoid starting the cache if it is defined but not started yet (example, the counter's internal cache is defined but it is only started when a counter is created)