Errors / checked October 9, 2026

job startup timeout

The error

The return_message of a failed row in cron.job_run_details (status failed), and a "cron job <id> job startup timeout" line in the Postgres log. Seen on Supabase Cron and any other Postgres with the pg_cron extension.

Quick fix

  1. Look at what else failed at the same minute: select jobid, start_time, status, return_message from cron.job_run_details order by start_time desc limit 50. Many jobs failing together points at the database being busy, not at one job.
  2. Spread jobs that share a schedule across different minutes, so they do not all open a connection at once. pg_cron runs up to cron.max_running_jobs (default 32) at the same time.
  3. If you run your own Postgres, check that the connection pg_cron makes is quick: point cron.host at a unix socket directory, or set cron.use_background_workers = on (and raise max_worker_processes) so jobs skip the client connection.

RunLater vs Supabase Cron

Why this happens

With its default settings pg_cron starts each job by opening a new client connection to the database (cron.host, localhost by default) and sending the command. If connecting and sending the command take longer than 10 seconds, it gives up and records "job startup timeout". The job's SQL never ran. Slow connections come from a database under heavy load, slow authentication, or many jobs starting in the same minute.

Source: pg_cron source (jobStartupTimeout, 10 s start deadline)

Other solutions

  • Make the job safe to rerun, so a run that never started can be run by hand.
  • Clean up cron.job_run_details on a schedule: pg_cron does not delete those rows itself, so the table grows with every run.
  • If the job only calls a URL with pg_net, a scheduler outside the database makes the call without using a Postgres connection or worker. RunLater calls the URL on your cron schedule, retries a failed call up to 10 times with exponential backoff, keeps each response 7 to 90 days by plan, and emails you when a run fails after its retries.

Check a schedule before you deploy it: the cron expression explainer shows the next 10 run times.

Frequently asked questions

Did my job run?

No. The timeout happens before the command is sent, so none of the job's SQL ran. The next scheduled run starts normally.

Can I change the 10 second limit?

No. It is a constant in pg_cron's source, not a setting.

What happens with background workers instead?

With cron.use_background_workers = on, a job that cannot get a worker slot within the same 10 seconds fails with "could not start background process; more details may be available in the server log", and the log says "out of background worker slots". Raise max_worker_processes.

Have the URL called on schedule, with retries

RunLater calls your URL on a cron schedule or at an exact time, in your timezone, retries failed calls with backoff and logs each response, so a timeout or a 5xx shows up in a log instead of disappearing.

Start a 14-day trialStarter is $15/month for 250 scheduled requests and 25,000 runs. Cancel before day 14 and you pay nothing.

Related errors