Run job in transaction about oban HOT 4 CLOSED

However, I haven’t found a way to save my data atomically with my job succeeding. There may be good reasons for this but would I can’t think of any. My use case is that I want to get some data from an external rest api and store the result in the database, could I do this within a transaction together with the job so that if it fails the data will never be stored unless the entire job succeeds?

There isn't anything built in that does this automatically, but you can easily handle it within your job. This example uses a private function from the Oban.Query module. All it is doing is setting the state to completed and recording a completed_at timestamp.

def perform(args, job) do
  changeset =
    args
    |> MyApp.RestAPI.get()
    |> MyApp.Record.changeset()

  MyApp.Repo.transaction(fn repo ->
    repo.insert!(changeset)
    Oban.Query.complete_job(Oban.config(), job)
  end)
end

If you happen to have multiple repos you can inject the transaction's repo into the config:

config = %{Oban.config() | repo: repo}

Alternately, if you'd rather not use Oban internals at all, you can make a changeset:

repo.update(Ecto.Changeset.change(job, state: "completed", completed_at: DateTime.utc_now()))

from oban.

Got it! So calling Oban.Query.complete_job is basically what Oban would do after my perform function and if I do it inside the function it won't cause any other issues?

from oban.

Exactly. Oban will mark complete a second time, but there isn’t any harm in it.

from oban.

Great, thanks for clarifying!

from oban.