Adding more counters to monitoring addons

[fjtirado]

Currently monitoring addons tracks the total number of completed instances (counter named kogito_process_instance_completed_total) and the total number of running instances (counter named kogito_process_instance_running_total). Although they are pretty useful to see how many processes are completed and how many processes are still running, it does not allow, from the total number of completed, to distinguish how many finished successfully and how many were aborted. The same is true for the ones running, it does not allow to distinguish how many are just executing and how many has suffered an error.

To resolve this, it would be nice to add, on top on the two mentioned counters, a counter for each relevant process status in the process engine:

• kogito_process_instance_sucessfully_completed: Numbers of process which status is completed. This counts the process that has completed its operation normally.
• kogito_process_instance_aborted: Numbers of process which status is aborted. They have been cancelled before finishing.
• kogito_process_instance_error: Numbers of process which status is error (waiting for user input, either to resolve the error or abort the process)
• kogito_process_instance_active: Numbers of process which status is active (either executing a service call or waiting for an event).

Note that the value of kogito_process_instance_completed_total is equal to the sum of kogito_process_instance_sucessfully_completed and kogito_process_instance_aborted and the value of kogito_process_instance_running_total is the sum of kogito_process_instance_error and kogito_process_instance_active

[fjtirado]

After a closer look to the micrometer implementation, there are some issues with the currrent proposal. Basically, the issue is that kogito_process_instance_running_total is a gauge and we will be difficult to keep the error count updated.
Therefore, an alternative proposal will be:

• Keep current kogito_process_instance_running_total gauge as it is.
• Add status tag (actually it already exist but with a wrong name) to kogito_process_instance_completed_total counter, so user can query for "completed" and "aborted" status (not need to have two different counters)
• Add a new kogito_process_instance_error counter which tracks the total number of errors that has occurred and do not decrease over time. Advantage of using a counter is that we can add an "error_message" type and user can calculate the total of error per process id and error type.