From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Dario Fumagalli <dfumagalli(at)tin(dot)it> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: Call closed: Almost happy ending (from "Data files became huge with no apparent reason" thread) |
Date: | 2002-09-03 14:32:47 |
Message-ID: | 5958.1031063567@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Dario Fumagalli <dfumagalli(at)tin(dot)it> writes:
> - Given that single user REINDEX [database name] FORCE in single user
> modedid rebuild the indexes but did not optimize them;
> - Given that a REINDEX INDEX [index name] did optimize and shrink a
> single index but increased the overall data size;
> - Given that a VACUUM VERBOSE ANALYZE somehow alters the overall
> database files size (seems to re-distribute the file space evenly across
> indexes);
There is no such thing as a "reindex that optimizes" versus "reindex
that doesn't optimize". A rebuild is a rebuild. I suspect that you
were fooled by out-of-date relpages statistics in pg_class, and/or
expecting REINDEX to rebuild indexes it won't.
VACUUM updates relpages to reflect current reality, but I'm not sure
whether REINDEX does.
Also, REINDEX at the database level only reindexes system-table indexes.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2002-09-03 14:35:13 | Re: DELETE SQL too slow. |
Previous Message | Gregory Seidman | 2002-09-03 14:30:01 | Re: getting y/n or 1/0 from bool fields |