There are a few instances where the release notes seem to indicate that the administrator should use pg_dump to upgrade a database so that improvements on btree can be available.
Here are they:
1.
>In new btree indexes, the maximum index entry length is
reduced by eight bytes, to improve handling of duplicate entries (Peter
Geoghegan)
2.
> This means that a REINDEX operation on an index pg_upgrade'd from a previous release could potentially fail.
>Improve performance and space utilization of btree indexes with many duplicates (Peter Geoghegan, Heikki Linnakangas)
>...
>Indexes pg_upgrade'd from previous releases will not have these benefits.
3.
>Allow multi-column btree indexes to be smaller (Peter Geoghegan, Heikki Linnakangas)
>...
>Indexes pg_upgrade'd from previous releases will not have these benefits.
My questions are:
1. Is this a current limitation of pg_upgrade that will be dealt afterwards?
2. Are we going to see more of such cases were pg_upgrade leaves the database incompatible with newer features.
3. What's the recommendation for administrators with databases that are too large to be upgraded with pg_dump?