Thanks Tom,
Dumped in Ubuntu 22.04_1 and restore attempted using Ubuntu 22.04.3
Editing the dump file to C.UTF8 didn't solve the problem.
The default for the database was en_AU.utf8 so I should have changed the
collation to that, but it was one field in one table of superseded data, so I
just erased the collation from that field.
Still no idea on how it came to be there.
Cheers
Tony
On Sunday 7 April 2024 10:35:44 AM ACST Tom Lane wrote:
> Tony Bazeley <tonyb@tonyb.id.au> writes:
> > I've a problem with restoring a cluster created with pg_dump_all from 14.8
> > ( pg_dumpall >pgall.out and then psql -f pgall.out postgres).
> > ...
> > Attempting to restore to postgresql-16 results in errors
> >
> > 2024-04-05 22:17:15.418 ACDT [6565] postgres@tonbaz ERROR: collation
> > "pg_catalog.C.UTF-8" for encoding "UTF8" does not exist at character 366
> >
> > I don't understand the class text COLLATE pg_catalog."C.UTF-8" syntax,
> > but
> > select * from pg_collation shows a C.UTF8 but no C.UTF-8
>
> I take it you are trying to restore onto a different OS platform with
> different locale naming conventions. The easiest way to deal with it
> probably is to edit the dump file and change "C.UTF-8" to "C.UTF8"
> everywhere. (Manually editing an 8G dump file might be no fun, but
> "sed" should make short work of it.)
>
> regards, tom lane