Thanks Scott. Could be my I/O as I am on 15k Raptor SATA drives with
RAID 1 only.
Anyway, reindexing happened fast (12 minutes) and things are now humming along.
How long should cluster take on a db that's about 5.5GB in size? Is it
worth doing on a production db? The db is running fast now anyway, so
the only reason I am even thinking about this is because of the
results of a VACUUM ANALYZE, which give me this:
---
INFO: "links": found 5427 removable, 8876599 nonremovable row
versions in 483102 pages
DETAIL: 96 dead row versions cannot be removed yet.
There were 2135991 unused item pointers.
235 pages contain useful free space.
---
My semi-intelligent question: in the output above, that figure of
"483,102" pages. Should this in any way affect my setting for
max_fsm_pages variable in pg.conf?