Hello,
I'm wondering about PostgreSQL always archiving the last replayed WAL
after getting out of recovery. This does not seem to be documented
anywhere, and this part of code is quite tricky to understand.
IMHO, PostgreSQL should not try to archive again WAL files coming from
the restore_command
op the dependent objects too.
Actually, the only way to "solve" this issue is to swap the relfilenode
columns between the old and the new index. I don't like this option very
much by the way but I'm very interested by your opinion on updating the
pg_class catalog this way.
Kind reg