[postgis-devel] ERROR: operator class "brin_geography_inclusion_ops" does not exist for access method "brin"

[Justin Pryzby]

On Sat, Oct 09, 2021 at 01:02:40PM -0400, Regina Obe wrote:
> > It seems very possible that postgis was not upgraded between versions as you
> > intended.  My usual procedure is to install the newest postgis for both the old
> > and new postgres version, and then run postgis_upgrade.sql, and then run
> > pg_upgrade.  But until recently I never ran ALTER EXTENSION postgis
> > UPGRADE, nor postgis_extensions_upgrade() - unless that's enforced, it
> > probably can't be assumed that someone did that.
> [Regina Obe] 
> Well in pre-PostGIS 3.0 if you never did that and were using extensions,
> then you never really upgraded because prior versions of postgis had
> distinct binaries for each minor.
> So for example postgis-2.5.so postgis-2.4.so and so on.  
> It was only in PostGIS 3.0 that we dropped the minor in the binary by
> default.  So perhaps you had all those lying around.

I don't think so.  My usual upgrade path from postgres A and postgis B to
postgres A+1 and postgis B+1 is to 1) uprade postgis to B+1 (so its C functions
would refer to the new postgis.so version), then pg_upgrade postgres A+1.
postgres A+1 would have never had postgis version B, and pg_upgrade would've
failed during its "check" phase if it didn't have B+1, and the old postgres
version would've been dysfunctional if the libraries for postgis B+1 wasn't
available.

It's entirely possible that my upgrades were incomplete, but not in any obvious
way.

> When did you switch from non-extension to extension version?

Do you mean CREATE EXTENSION rather than \i postgis.sql ?  I suppose the answer
is that I haven't yet switched, or finished switching.

I'd prefer to use "CREATE EXTENSION", but it looks like on many of our
instances, postgis is not listed in psql \dx, and it was never clear to me how
to fix it.  We only have postgis for 2 columns used by mapscript, so if that
works, it's good enough for us.

-- 
Justin