We are using WildFly, but I don’t think there’s an appreciable difference in this context. One easy way to determine if Camunda is working is to write a simple process and run it. If the process can print (e.g. System.out.println(“I’m working”) in a Groovy script listener) to a server log, then you would see evidence there of Camunda “working”. In your case with a cluster, this might be more difficult unless you can consolidate server logs.
I’ve built various utility processes that test basic function, external services we’ve written, etc.
The most basic way to tell if Camunda is up is if the REST API responds. For example if you use curl on one of the hosts, you might execute the following and you should get some form of response (assuming at least one process is deployed):
This assumes you don’t have any authentication requirements on the REST API. If you do, then you could use something like this:
These and other methods you can find in the REST API documentation could be used to determine if it’s working.
With respect to the database, while Oracle may support READ COMMITTED isolation, are you actually using it? Is your Oracle database a cluster of any sort? The last time I checked, the only supported database cluster architecture was MariaDB Galera cluster, configured for READ COMMITTED isolation. The only advantage of that cluster is some resiliency in the database, but you pay a price for that.