Re: recovering from "found xmin ... from before relfrozenxid ..."
Stephen Frost <sfrost@snowman.net>
From: Stephen Frost <sfrost@snowman.net>
To: Magnus Hagander <magnus@hagander.net>
Cc: Robert Haas <robertmhaas@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Andres Freund <andres@anarazel.de>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2020-07-14T14:36:25Z
Lists: pgsql-hackers
Greetings, * Magnus Hagander (magnus@hagander.net) wrote: > On Tue, Jul 14, 2020 at 4:09 PM Robert Haas <robertmhaas@gmail.com> wrote: > > On Tue, Jul 14, 2020 at 8:25 AM Magnus Hagander <magnus@hagander.net> > > wrote: > > > I don't think that it necessarily has to be. As long as we're talking > > about adding something and not actually changing their existing packages, > > getting this into both yum and apt shouldn't be *that* hard, if it's > > coordinated well with Christoph and Devrim (obviously that's based on my > > experience and they will have to give a more complete answer themselves). > > It would be a lot more complicated if it involved changing an existing > > package. > > > > I mean, you presumably could not move pg_resetwal to this new package > > in existing branches, right? > > Probably and eventually. But that can be done for 14+ (or 13+ depending on > how "done" the packaging is there -- we should just make sure that hits the > biggest platform in the same release). Considering we just got rid of the -contrib independent package on at least Debian-based systems, it doesn't really seem likely that the packagers are going to be anxious to create a new one- they are not without costs. Also, in such dire straits as this thread is contemplating, I would think we'd *really* like to have access to these tools with as small an amount of change as absolutely possible to the system: what if pg_extension itself got munged and we aren't able to install this new contrib module, for example? I would suggest that, instead, we make this part of core, but have it be in a relatively clearly marked special schema that isn't part of search_path by default- eg: pg_hacks, or pg_dirty_hands (I kinda like the latter though it seems a bit unprofessional for us). I'd also certainly be in support of having a contrib module with the same functions that's independent from core and available and able to be installed on pre-v14 systems. I'd further support having another repo that's "core maintained" or however we want to phrase it which includes this proposed module (and possibly all of contrib) and which has a different release cadence and requirements for what gets into it, has its own packages, etc. Thanks, Stephen
Commits
-
Fix wrong data table horizon computation during backend startup.
- 1c7675a7a426 14.0 landed
-
Centralize horizon determination for temp tables, fixing bug due to skew.
- 94bc27b57680 14.0 landed
-
pg_surgery: Try to stabilize regression tests.
- 0811f766fd74 14.0 landed
-
New contrib module, pg_surgery, with heap surgery functions.
- 34a947ca13e5 14.0 landed
-
Set cutoff xmin more aggressively when vacuuming a temporary table.
- a7212be8b9e0 14.0 cited
-
snapshot scalability: Don't compute global horizons while building snapshots.
- dc7420c2c927 14.0 cited
-
Introduce vacuum errcontext to display additional information.
- b61d161c1463 13.0 cited