For PG we want to execute some checks on all DBs in each instance. Some our customers have hundreds DBs in one instance. With current implementation check may be started simultaneously for all DBs which may create spike load on DBMS. We should limit amount of concurrent jobs and a amount of opened connections for DBMS instance.
Out of scope:
1) Use one runner for jobs and actions
2) Use semaphore to limit total amount of jobs/actions
3) Use semaphore to limit amount of connections per DBMS instance
How to test:
Just check that action and jobs works. For example create some backup with PMM, open QAN and invoke explain for some query there, etc...