> On May 20, 2019, at 11:12 AM, Tom Lane <t...@sss.pgh.pa.us> wrote: > > Hmm, so not a deadlock (plus you're running a version that has that > bug fixed). It sounds like you might have hit an actual bug, perhaps > one that causes it to endlessly re-split an index page, or something > along that line. But it's going to be hard to find the cause without > a test case. I don't suppose you can anonymize your data to the > point where you'd have a publishable test case?
That would take some time, thankfully it seems reproducible with a few million rows. I’ve moved the files around a little bit (not aggressively) and it still stalled, so I don’t think it’s something specific to the data. But we’ll have to play with it and see. Thanks.
