1---
2title: Use SKIP LOCKED for Non-Blocking Queue Processing
3impact: MEDIUM-HIGH
4impactDescription: 10x throughput for worker queues
5tags: skip-locked, queue, workers, concurrency
6---
7 
8## Use SKIP LOCKED for Non-Blocking Queue Processing
9 
10When multiple workers process a queue, SKIP LOCKED allows workers to process different rows without waiting.
11 
12**Incorrect (workers block each other):**
13 
14```sql
15-- Worker 1 and Worker 2 both try to get next job
16begin;
17select * from jobs where status = 'pending' order by created_at limit 1 for update;
18-- Worker 2 waits for Worker 1's lock to release!
19```
20 
21**Correct (SKIP LOCKED for parallel processing):**
22 
23```sql
24-- Each worker skips locked rows and gets the next available
25begin;
26select * from jobs
27where status = 'pending'
28order by created_at
29limit 1
30for update skip locked;
31 
32-- Worker 1 gets job 1, Worker 2 gets job 2 (no waiting)
33 
34update jobs set status = 'processing' where id = $1;
35commit;
36```
37 
38Complete queue pattern:
39 
40```sql
41-- Atomic claim-and-update in one statement
42update jobs
43set status = 'processing', worker_id = $1, started_at = now()
44where id = (
45  select id from jobs
46  where status = 'pending'
47  order by created_at
48  limit 1
49  for update skip locked
50)
51returning *;
52```
53 
54Reference: [SELECT FOR UPDATE SKIP LOCKED](https://www.postgresql.org/docs/current/sql-select.html#SQL-FOR-UPDATE-SHARE)
55