William Yu <[EMAIL PROTECTED]> writes:
> It seems that upon dump & restore, UPPER indexes either aren't recreated
> correctly or not listed somewhere the query analyzer can know it exist.
Seems unlikely. Perhaps you forgot to ANALYZE after reloading?
regards, tom lane
William Yu <[EMAIL PROTECTED]> writes:
> Index Scan using idx_finvendors_name on fin_vendors (cost=0.00..4.01
> rows=1 width=600) (actual time=0.029..0.036 rows=2 loops=1)
> Index Cond: ((name >= 'NBC'::bpchar) AND (name < 'NBD'::bpchar))
> Filter: (name ~~ 'NBC%'::text)
Hmm. Apparent
Certainly did analyze. Here's the query plans. Note the non-UPPER query
uses an indexscan just fine.
INFO: analyzing "public.fin_vendors"
INFO: "fin_vendors": 4207 pages, 3000 rows sampled, 63063 estimated
total rows
ANALYZE
talisman=# explain analyze select * from fin_vendors where name like
It seems that upon dump & restore, UPPER indexes either aren't recreated
correctly or not listed somewhere the query analyzer can know it exist.
I've encountered first encountered this problem doing an upgrade to
7.3.7 to 7.4.6. I again encountered this program replicating a server
(same 7.4.6
