After importing data in PostgreSQL, duplicate key value violates unique constraint

Question

I recently migrated my rails app to PostgreSQL in order to take advantage of fulltext search.

Since the migration coincided with moving to a new webhost, the steps for m

Answer 1

You can automate wildplasser's solution so that all sequences are synchronized with the current maximum value of their associated column:

do
$block$
declare 
  r        record;
  stmt     text;
  max_id   integer;
begin
  for r in (
              select *
              from (
                select table_schema, 
                       table_name, 
                       column_name, 
                       pg_get_serial_sequence(table_schema||'.'||table_name, column_name) as col_sequence
                from information_schema.columns
                where table_schema not in ('pg_catalog', 'information_schema')
              ) t
              where col_sequence is not null
        ) 
  loop
    stmt := 'select coalesce(max('||r.column_name||'), 0) + 1 from '||r.table_schema||'.'||r.table_name;
    execute stmt into max_id;
    raise notice 'Next ID for %.%.% is %', r.table_schema, r.table_name, r.column_name, max_id;
    perform setval(r.col_sequence, max_id); 
  end loop;
end;
$block$

Note that this will only work if the columns have been defined as serial, bigserial or have been made the "owner" of a sequence.

Answer 2

In Rails you can use the command

ActiveRecord::Base.connection.reset_pk_sequence!('users')

to bring the primary key index for the User table in sync again.

Answer 3

If the schema contains serial or sequence columns, you should reset these to the max value that occurs in the corresponding column. (normally you should not import the serials from a file, but give them the freedom to autoincrement.)

For all imported tables you should identify the sequence fields and run the following code on them. (substitute your schema name for "sch", your table name for "mytable" and your id column name for "id")

WITH mx AS ( SELECT MAX(id) AS id FROM sch.mytable)
SELECT setval('sch.mytable_id_seq', mx.id) AS curseq
FROM mx
        ;

Answer 4

Another way is remove the primary key (id) from the columns (or don't dump the id). So your data would look like

INSERT INTO book (name, price) VALUES ('Alchemist' , 10);

instead of

INSERT INTO book (id, name, price) VALUES (1 , 'Alchemist' , 10);

This way, you won't have to reset the primary key after loading initial data