Back-filling a nullable column with sequence

As part of an effort to create a primary key for a large table in production, I've created a nullable id column, an auto-incrementing sequence and connected the two:

-- Add nullable column id
ALTER TABLE my_table ADD COLUMN IF NOT EXISTS id BIGINT;

-- Create a sequence
CREATE SEQUENCE my_table_id_seq;

-- Associate the sequence with the table
ALTER SEQUENCE my_table_id_seq OWNED BY my_table.id;

-- Use the sequence for new rows in this table
ALTER TABLE my_table ALTER COLUMN id SET DEFAULT nextval('my_table_id_seq');

After this was executed all new rows indeed getting incrementing values on the id column. Now I need to back-fill all previously existing rows with values from the sequence.

I've had some concerns:

• I suppose that for each row where the value of id is NULL I need to use the next value from the sequence as well as increment the index, in order to avoid using the same value twice.
• What if there will be concurrent inserts to this table - can this create a race condition and cause more than one row with the same id value?
• Updating all rows in one go might take a long time - instead, I intend to do the update in batches of 10K rows or so by previously selecting rows for update in a CTE.

What would be the best way to use the sequence to back fill all rows where id IS NULL while staying safe vs. concurrent inserts?

You can update the current records, using nextval()

UPDATE my_table 
SET id = nextval('my_table_id_seq') 
WHERE id IS NULL;

Maintenance of the index is done by the database and you don't have issues with concurrency. Just don't expect any order in your numbers, these are just numbers without any additional meaning.