What is the best index for this delayed_job request for postgres?

Question

delayed_job regularly executes this request:

SELECT "delayed_jobs".* FROM "delayed_jobs" WHERE ((run_at <= '2012-05-23 15:16:43.180810' AND (locked_at IS NULL OR locked_at < '2012-05-23 11:16:43.180841') OR locked_by = 'host:foo pid:1') AND failed_at IS NULL) ORDER BY priority ASC, run_at ASC LIMIT 5 

My logs on my rather large database computer report that it takes a quarter of a second to start. I could just add some indexes to all the columns that were selected, but I probably get more performance from an index with multiple columns.

What is the best index for multiple columns I can do for this query? Are there any tools that can calculate this for me?

Update

postgres version: 9.1.3

one existing index: priority, run_at (named "delayed_jobs_priority")

from explain analyze :

 Limit (cost=0.00..219.65 rows=5 width=1154) (actual time=0.727..0.727 rows=0 loops=1) -> Index Scan using delayed_jobs_priority on delayed_jobs (cost=0.00..351.43 rows=8 width=1154) (actual time=0.725..0.725 rows=0 loops=1) Filter: ((failed_at IS NULL) AND (((run_at <= '2012-05-23 18:11:03.980113'::timestamp without time zone) AND ((locked_at IS NULL) OR (locked_at < '2012-05-23 14:11:03.98014'::timestamp without time zone))) OR ((locked_by)::text = 'host:foo pid:1'::text))) Total runtime: 0.754 ms (4 rows) 

Answer 1

Since you have a LIMIT , it is possible that you need an ordering index instead of a filtering one on (priority, run_at) .

What is the percentage of records in your table that satisfy the WHERE ?

Answer 2

I do not think that in this case the multi-column index is very useful. Use multiple indexes on the same column.