Created attachment 832060 [details] Patch described We are doing in Availmanager.mergeAvailInNewTx for ( Avail a: incoming ) { latest = db.find ( a ); ... latestRA = db.find(ResourceAvail.class, a.resoure.id); } So with a report of 10k availabilities we fire >20k db requests and round trips to the database to just get the current state of affairs. We should pull that out to reduce the number of round trips to the DB. See patch that illustrates what I mean (we should use probably use named queries, that the JPA subsystem can pre-compile into prepared statements).
See also Bug 1032192 that increases batch sizes for writing the (changed) data back to the DB.
I agree that we could avoid some db fetches here. I'm not sure how significant a boost we'll see but this is a critical code section for scalability and perhaps any boost is worth a change. I've reviewed the patch and will apply similar changes. Although I don't think the patch as it stands is quite right in its approach to the previous exception handling cases.
master commit 544dc80ce4c22e399d3ca607108896feb6ebd5b6 Author: Jay Shaughnessy <jshaughn> Date: Wed Dec 4 13:36:49 2013 -0500 Optimization work based on BZ-supplied patch. For a batch of Availabilities being processed, pull latest availabilities for the resources in one DB round trip as opposed to once for each resource. Still beiong careful to repair issues when no, or multiple, latest entries are found. master commit 0beeff7ee267456050c572124b5e4103bd377c1c Author: Jay Shaughnessy <jshaughn> Date: Thu Dec 5 11:37:30 2013 -0500 Part 2 - batch the pull of ResourceAvailability records that need to be updated due to changed availability. Pull only the necessary records. Test Notes: This is a performance optimization. Testing here is implicit, given no regressions in existing testing, this can be marked verified.
master commit b84dcfa9e3f141d2431665b592cab9c95655a0e7 Author: Jay Shaughnessy <jshaughn> Date: Fri Dec 6 16:01:03 2013 -0500 A fix in the new impl to handle multiple updates for the same resource in the same report. In general an avail report should only have one update for one resource, but the potential for building up reports looms in the future, and we do have this scenario in test code.
Bulk closing of 4.10 issues. If an issue is not solved for you, please open a new BZ (or clone the existing one) with a version designator of 4.10.