On Wed, Dec 20, 2017 at 12:46 AM, Robert Haas <robertmh...@gmail.com> wrote: > On Tue, Dec 19, 2017 at 5:27 AM, Feike Steenbergen > <feikesteenber...@gmail.com> wrote: >> While doing some wraparound debugging, I saw the hint regarding upcoming >> wraparound did not include the problem of having a stale replication >> slot (which I'm actually using to force wraparound issues). >> >> I remember a few discussions where a stale replication slot was actually the >> culprit in these situations. >> >> Something like the attached maybe? > > +1.
+1. prepare_transaction.sgml has a "Caution" block mentioning that it is unwise to keep 2PC transactions unfinished for a too-long time as it interferes with VACUUM. In doc/src/sgml/logicaldecoding.sgml, it would be nice to add the a similar caution notice about replication slots so as users can get be warned before a wraparound shows up. -- Michael
