I'm worried about the refresh calls here. In large requests (such as those for upgrades), there could be 10's of 1000's of HRCs. Calling refresh on each one will incur DB hits and potentially cause the web client to timeout on things like Pause/Downgrade.
I think we need to do some more investigation of this code before it gets checked in. ' In general, I think that requiring Java-level locks around JPA means that something is wrong with our JPA design and that's where it should be fixed. `refresh()` is dangerous in that it makes it look like we're fixing a problem, yet we're introducing much larger ones that only customers see. [ Full content available at: https://github.com/apache/ambari/pull/2411 ] This message was relayed via gitbox.apache.org for [email protected]
