On Fri, Apr 05, 2019 at 09:41:58AM +0200, Michael Banck wrote: > Is there a good reason why Postgres doesn't just issue a CHECKPOINT > after promote itself? After all, this seems to be about making the > control file having the proper content, which sounds like a good thing > to have in general.
The startup process requests a checkpoint since 9.3, and before that it was doing the checkpoint by itself (grep for fast_promoted and RequestCheckpoint() around line 7579 in xlog.c). This allows the recovery to finish much faster.
> Could this be a problem for anything else besides pg_rewind?
Not that I know of, at least not in the tree.
> This looks like a needless footgun waiting to happen, and just > documenting it in pg_rewind's notes section looks a bit too hidden to me > (but is certainly an improvement).
We had a couple of reports on the matter over the past years. Perhaps we could use a big fat warning but that feels a bit overdoing it.
A related question is, could we (for 12+) actually make the problem go away? As in, can we detect the state and just have pg_rewind issue the checkpoint as needed?