Hundreds of indexes created automatically - sql

Question

I created a table which is used for storing information relative to resources of different kind, the table description prints 250 indexes for each foreign key, is this behavior normal?
The table contains only 50 rows.
CREATE TABLE IF NOT EXISTS resource
(
id serial PRIMARY KEY,
kind resource_kind NOT NULL,
author_id int REFERENCES user(id) ON DELETE SET NULL,
---uri text NOT NULL,
creation_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
modification_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
message_id int references message,
track_id int references track,
location_id int references track,
fileupload_id int references upload
check (
(
(message_id is not null)::integer +
(track_id is not null)::integer +
(location_id is not null)::integer +
(fileupload_id is not null)::integer
) = 1
)
);
create unique index on resource (message_id) where message_id is not null;
create unique index on resource (track_id) where track_id is not null;
create unique index on resource (location_id) where location_id is not null;
create unique index on resource (fileupload_id) where fileupload_id is not null;
And this is the output of "\d resource"
Colonna | Tipo | Ordinamento | | Default
-------------------+-----------------------------+-------------+-----------------+------------------------------------------------
id | integer | | not null | nextval('resource_id_seq'::regclass)
kind | resource_kind | | not null |
author_id | integer | | |
creation_time | timestamp without time zone | | not null | CURRENT_TIMESTAMP
modification_time | timestamp without time zone | | not null | CURRENT_TIMESTAMP
message_id | integer | | |
track_id | integer | | |
location_id | integer | | |
fileupload_id | integer | | |
Indici:
"resource_pkey" PRIMARY KEY, btree (id)
"resource_fileupload_id_idx" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx1" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx10" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx100" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx101" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx102" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx103" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx104" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx105" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx106" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
"resource_fileupload_id_idx107" UNIQUE, btree (fileupload_id) WHERE fileupload_id IS NOT NULL
...
And then the outputs goes with hundreds of indexes for each foreign key.

Answer 1

No, that is not normal.
Somebody or some software must have created the indexes.
Remove them all except for one, because they use space and will make data modifications unbearably slow.

Answer 2

Next time you could put all the constraints inside the DDL for the table, making it conditional on NOT EXISTS (this will also create unique names for the indexes) :
CREATE TABLE IF NOT EXISTS resource
(
id serial PRIMARY KEY
, kind INTEGER NOT NULL -- resource_kind NOT NULL
, author_id INTEGER -- REFERENCES user(id) ON DELETE SET NULL
-- , uri text NOT NULL
, creation_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
, modification_time TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP
, message_id INTEGER -- references message
, track_id INTEGER -- references track
, location_id INTEGER -- references track
, fileupload_id INTEGER -- references upload
, unique (message_id)
, unique (track_id)
, unique (location_id)
, unique (fileupload_id)
, check (
(
(message_id is not null)::integer
+ (track_id is not null)::integer
+ (location_id is not null)::integer
+ (fileupload_id is not null)::integer
) = 1
)
);
-- check it
\d resource