Alvaro Herrera <alvherre@atentus.com> writes:
> But at least in 7.1 REINDEX has to be run under stand-alone Postgres, or
> so say the docs:
The docs are not very well phrased (I've just made a note to fix that).
If you are trying to recover from a corrupted system index then indeed
you have to do all the pushups mentioned, because you don't want the
backend trying to use the broken index along the way:
> REINDEX is used to recover corrupted system indexes. In order to run
> REINDEX command, postmaster must be shut down and stand-alone Postgres
> should be started instead with options -O and -P (an option to ignore
> system indexes).
If you are trying to optimize an index on a user table, you can just do
REINDEX INDEX or REINDEX TABLE without the funny stuff.
> Whereas the DROP INDEX/CREATE INDEX method works on-the-fly. You may end
> up running a seqscan somewhere, but that's not so big a problem, IMHO.
True, that way doesn't lock out concurrent readers of the table while
the index is rebuilt.
regards, tom lane
