projects / users / hanada / postgres.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b25433d) | patch
Fix logic in lazy vacuum to decide if it's worth trying to truncate the heap.
authorHeikki Linnakangas <heikki.linnakangas@iki.fi>
Tue, 6 Jan 2009 14:55:37 +0000 (14:55 +0000)
committerHeikki Linnakangas <heikki.linnakangas@iki.fi>
Tue, 6 Jan 2009 14:55:37 +0000 (14:55 +0000)
commit7ffe657225ad38b912688769ffb58e3d450ef17d
tree6d56f1ca4e3f0c8107288e688ab0507f14ff0e78tree
parentb25433da5d826e2d83c8f026ed81d5870f511f0fcommit | diff
Fix logic in lazy vacuum to decide if it's worth trying to truncate the heap.
If the table was smaller than REL_TRUNCATE_FRACTION (= 16) pages, we always
tried to acquire AccessExclusiveLock on it even if there was no empty pages
at the end.

Report by Simon Riggs. Back-patch all the way to 7.4.
src/backend/commands/vacuumlazy.c diff | blob | blame | history
Hanada's development tree.