Hannu, > If it is a sequence of > > 1. ordinary VACUUM (it can't run out of FSM anymore, no?) > 2. a process moving live tuples from end (using reverse seqscan) to free > space found scanning in first-to-last direction, either one tuple at a > time or one page at a time, until the two scans meet > 3. another ordinary VACUUM to actually reclaim the free space > > 4. repeat a few times so that tuples at the end of relation (for > whatever reason) added while doing 1-3 are also moved towards beginning
Sounds good, you want to code it for 8.5? I could actually see two tools, one VACUUM FULL CONCURRENTLY and one VACUUM REWRITE. The first would be "in place" and the second would be "fast". Both should work better with HS than current VF does. -- Josh Berkus PostgreSQL Experts Inc. www.pgexperts.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
