Image job statuses for a progress UI: queued is not failed

A progress UI for Sume image jobs needs five labels, one for each job status: queued, processing, completed, failed, and canceled. The one people get wrong is queued, a normal accepted state that means the job is waiting for a workspace concurrency slot, not that something broke.

Sume's docs say to store the job id, poll status with backoff, and fetch the result only when the job reports result_ready: true or status: completed.

What do the statuses mean?

Three of the five are terminal. After a terminal status, stop polling.

How often should I poll?

Use exponential backoff and stop on a terminal status. The SDK docs say the status payload's next_poll_after_seconds wins over your own interval when it asks for a longer gap, so read it if present. Reading the events timeline (job.created, job.queued, job.started, job.completed) helps explain a long wait.

What should the failed state show?

Failed jobs carry public error metadata: category, stage, retryability, retry-after seconds, public reason, and next action. Show the next action, not a raw error. Validation means fix the input; quota means add funds or lower cost; queue means retry later with the same idempotency key. Do not resubmit a paid request just because the UI lost the connection; poll the job id instead.

Images that run past a minute are covered in image job running past 50 seconds. The status list is in the Jobs and results docs.