a better query to delete from hashtags

old query:

Delete on hashtags  (cost=5089.81..5521.63 rows=6160 width=18)
   ->  Hash Semi Join  (cost=5089.81..5521.63 rows=6160 width=18)
         Hash Cond: (hashtags.id = ht.id)
         ->  Seq Scan on hashtags  (cost=0.00..317.28 rows=17528 width=14)
         ->  Hash  (cost=5012.81..5012.81 rows=6160 width=20)
               ->  Merge Anti Join  (cost=0.70..5012.81 rows=6160 width=20)
                     Merge Cond: (ht.id = hto.hashtag_id)
                     ->  Index Scan using hashtags_pkey on hashtags ht  (cost=0.29..610.53 rows=17528 width=14)
                     ->  Index Scan using hashtags_objects_pkey on hashtags_objects hto  (cost=0.42..3506.48 rows=68158 width=14)

new query:

Delete on hashtags ht  (cost=0.70..5012.81 rows=6160 width=12)
   ->  Merge Anti Join  (cost=0.70..5012.81 rows=6160 width=12)
         Merge Cond: (ht.id = hto.hashtag_id)
         ->  Index Scan using hashtags_pkey on hashtags ht  (cost=0.29..610.53 rows=17528 width=14)
         ->  Index Scan using hashtags_objects_pkey on hashtags_objects hto  (cost=0.42..3506.48 rows=68158 width=14)
This commit is contained in:
faried nawaz 2021-04-21 03:52:32 +05:00
parent a0c9a2b4cc
commit 5be9d13981
No known key found for this signature in database
GPG key ID: C8EBC3669D3539B4

View file

@ -97,12 +97,10 @@ def run(["prune_objects" | args]) do
|> Repo.delete_all(timeout: :infinity)
prune_hashtags_query = """
delete from hashtags
where id in (
select id from hashtags as ht
left join hashtags_objects as hto
on hto.hashtag_id = ht.id
where hto.hashtag_id is null)
delete from hashtags as ht
where not exists (
select 1 from hashtags_objects hto
where ht.id = hto.hashtag_id)
"""
Repo.query(prune_hashtags_query)