Re: VACUUM FULL doesn't reduce table size

David G Johnston <david(dot)g(dot)johnston(at)gmail(dot)com> writes:
> I have no experience here but given recent versions rewrite the table the
> vacuum verbose output shown seems unusual.

"vacuum verbose output shown?" There wasn't any.

[ digs about ... ]

Oh. The version of the message that nabble sent to the postgresql lists
was missing vital parts of what got posted at nabble: compare
http://www.postgresql.org/message-id/1425639585904-5840782.post@n5.nabble.com
http://postgresql.nabble.com/VACUUM-FULL-doesn-t-reduce-table-size-td5840782.html

I've seen that before. I'm about ready to propose that we flat out ban
messages from nabble to the PG lists; I'm tired of them wasting our time
with this sort of BS.

(FWIW, the output shown on nabble doesn't look materially different from
what I see in HEAD. It also proves positively that the OP has uncommitted
transactions blocking cleanup of known-dead rows. But we were just
guessing at that over here.)

regards, tom lane