Would I be correct in assuming that the following two indexes are
completely redundant except for the fact that one complains about
uniqueness constraint violations and the other does not?
Or is there are legitimate use for having BOTH indexes?
I'm trying to figure out if it's okay to delete the non-unique index.
(I have a bunch of tables suffering this malady from some problematic
application code).
Table "public.erf"
Column | Type | Modifiers
--------+---------+-----------
rid | integer | not null
cid | integer | not null
Indexes: erf_rid_key unique btree (rid),
erf_rid_idx btree (rid)
Index "public.erf_rid_idx"
Column | Type
--------+---------
rid | integer
btree, for table "public.erf"
Index "public.erf_rid_key"
Column | Type
--------+---------
rid | integer
unique, btree, for table "public.erf"
