Hi all! Are there ways to speedy update of hundreds of millions of rows in comressed hypertable. I tried approach described in article. But the update affected only a small part of the values.
I checked the update on one compressed chunk only.
If I understand correctly, the values in the field were updated for those records that were decompressed into the parent chunk from the compressed child chunk before I ran the script to update the values. That is, those that are still in the compressed chunk were not updated (but I don’t know how to check it).
The following steps work correctly (it was suggested to me by my wife, who is not a developer Image may be NSFW.
Clik here to view.
): decompress the chunk, update the values, compress the chunk. In the absence of any other option, this one is a working one. But it seems to me that there should be another way to update all the values in the field in a compressed chunk without decompressing the chunk.
My script of update:
DO $$
DECLARE
v_limit INTEGER := 3000;
v_row_id_del_to INTEGER;
v_exit BOOLEAN := FALSE;
BEGIN
LOOP
WITH
cte AS (
SELECT old_value, new_value
FROM tmp
ORDER BY row_id
LIMIT v_limit
)
UPDATE _timescaledb_internal.<uncompressed_chunk_name> AS tgt
SET field_to_update = cte.new_value
FROM cte
WHERE tgt.field_to_update = cte.old_value
;
COMMIT;
-- ..... other code
END LOOP;
END $$;
2 posts - 2 participants
