What is this issue?
The lock_timeout setting specifies how long to wait for a lock before
giving up. Without a timeout:
- Queries wait forever for locked resources
- DDL operations (ALTER TABLE) block until locks are available
- Blocking chains pile up waiting sessions
- Application threads are tied up waiting
A lock_timeout allows "fail fast" behavior where blocked operations
error immediately, allowing the application to retry or handle gracefully.
Why it matters
Cascading Blocks
Waiting queries pile up, exhausting connections
Application Hangs
Threads/processes stuck waiting for locks
DDL Risk
Schema changes can wait hours for locks
Poor User Experience
Users see hanging requests instead of errors
How PG Pilot detects it
```sql SELECT name, setting, unit FROM pg_settings WHERE name = 'lock_timeout'; ```
How to fix it
1
Set a reasonable default
30 seconds to 1 minute is common:
ALTER SYSTEM SET lock_timeout = '30s';
SELECT pg_reload_conf();
2
Use shorter timeouts for DDL
Fail fast for schema changes:
SET lock_timeout = '5s';
ALTER TABLE users ADD COLUMN new_col text;
RESET lock_timeout;
3
Application retry logic
Handle lock timeouts gracefully:
begin
execute_query
rescue ActiveRecord::LockWaitTimeout
sleep(rand * 2)
retry
endPrevention
- Set a default lock_timeout
- Use short timeouts for DDL
- Implement retry logic in applications
- Monitor blocking with PG Pilot