ScraperData

Async tasks

Async tasks

Add a pingback URL to a search to queue it instead of waiting for the answer. Send 100 at once on a plan of 40, and all 100 are accepted.

How it works

  1. Queue a search: a normal search with a pingback_url. It returns at once with 202 and a task id.
  2. We run it as search capacity frees up, up to your plan's concurrency at a time, in turns with other customers.
  3. When it ends we call your URL with the id and the status: the pingback.
  4. You read the answer with GET /v1/tasks/{id}.

Why queue

Live searches above your plan's concurrency are refused with 429. Queued searches are never refused for that reason, so you can send a whole batch and collect the answers as they come.

Good to know

  • Credits are held when a search is queued and charged only for the pages that come back. A failed search costs nothing.
  • Searches usually start within seconds. A failing search is tried up to 3 times.
  • A search that cannot run within an hour ends with task_expired, and its credits go back.
  • Up to 10,000 of your searches can wait or run at once; more get 429 queue_full.
  • Answers are kept for 7 days.