Whit Armstrong <armstrong.whit@gmail.com> writes:
> getting a few "page xxx is uninitialized --- fixing" warnings in the
> vacuum output, but seems like this should be expected since I killed
> an insert in progress.
Yeah, that isn't too surprising. When a table runs out of space the
first thing we do is extend it with zeroed pages, mainly to make sure
that the disk space is available. The next thing that will hit disk
is WAL records about the insertions, and finally the updated pages
themselves (overwriting the zeroes). A forced database crash before
the WAL gets written leads to some all-zero pages left behind on disk.
They're harmless, and VACUUM knows enough to clean them up and make
them useful (but it emits a WARNING just to be chatty).
regards, tom lane