windmill-labs/windmill · error

could not create job token

Error message

could not create job token

What it means

Fires in the monitor's ephemeral-job path when create_token_for_owner fails to mint the short-lived job token (label from ephemeral_script_token_label, expiry from job_token_expiry_secs) for a queued job. The underlying cause is a database failure during token insertion/lookup — the queued job itself could not be turned into an executable job.

Source

Thrown at backend/src/monitor.rs:5592

        if let Err(e) = job {
            tracing::error!("Error getting queued job: {:?}", e);
            continue;
        }
        if let Some(job) = job.unwrap() {
            let label = ephemeral_script_token_label(&job.permissioned_as, &job.created_by);
            let token = create_token_for_owner(
                &db,
                &job.workspace_id,
                &job.permissioned_as,
                &label,
                job_token_expiry_secs(&db, &job.workspace_id).await,
                &job.permissioned_as_email,
                &job.id,
                None,
                Some(format!("handle_zombie_jobs")),
            )
            .await
            .expect("could not create job token");

            let client = AuthedClient::new(
                base_internal_url.to_string(),
                job.workspace_id.to_string(),
                token,
                None,
            );

            let error_message = format!(
                "Job timed out after no ping from job since {} (ZOMBIE_JOB_TIMEOUT: {}, reason: {:?}).\nThis likely means that the job died on worker {}, OOM are a common reason for worker crashes.\nCheck the workers around the time of the last ping and the exit code if any.",
                job.last_ping.unwrap_or_default(),
                *ZOMBIE_JOB_TIMEOUT,
                error_kind.to_string(),
                job.worker.clone().unwrap_or_default(),
            );
            let memory_peak = job.memory_peak.unwrap_or(0);
            let (_, killpill_rx_never_used) = KillpillSender::new(1);
            let _ = handle_job_error(

View on GitHub (pinned to e474e8803c)

Solutions

1. Check database connectivity and health from the worker/monitor
2. Look for DB constraint or capacity errors around the tokens table at that timestamp
3. Restart the worker if the DB was restarted underneath it

When it happens

Trigger: Thrown at backend/src/monitor.rs:5592 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of windmill-labs/windmill@e474e8803c (2026-09-03). Data as JSON: /api/errors/487e6bdba91fa740. Report an issue: GitHub.