Arnt Gulbrandsen wrote: >That that that's what I was running when Abhijit and I tried to do >the same thing, and failed. IIRC, the code ended up paying a >substantial part of the whole-vacuum cost for each chunk. At the end >we gave up, ran the whole thing, and made -a optional.
Ah, ok. Well, maybe PostgreSQL 9.x got cleverer in the light of a LIMIT. Incidentally, the -a option is broken in my version (either by me, or it always has been). -- Stephen.
