Comment 2 for bug 590992

Query plans are slightly different on lpnet vs staging, which is causing the timeout. See: https://pastebin.canonical.com/33112/

Presumably, something in the difference in data is causing postgresql to choose differently and causing the slowdown. I don't think there's anything new in this code path to suddenly have caused this. I'll have to dig more later, but wanted to update the bug in case others poke tonight or tomorrow AM before I do.