Comment 2 for bug 1738488

I should add some more information relevant to my previous comment:

If I wait a couple of minutes and try that patron's payment history again, I usually get it. I believe this is because the query has finally finished and the results have been cached (somewhere).

I think the main reason for the huge discrepancy in run time between production and the other servers is not so much because of production's better hardware. We have set join_collapse_limit to 9 on production. That settings aids with the performance of queries with massive joins.

As a final thing worth noting, this does not happen in the XUL client. I've not looked, but I assume that the web staff client is doing something very different from the XUL client with respect to patron payment history.