Allow Jobs to run in sequence

[amitaibu]

I have a demo "Seat reservation" app that allows students to reserve a seat.

When students asks for a seats, a Reservation record is saved, and a unique seat number should be later assigned to it via a Job.
See it in action.

When I try to register many students at once, I get duplicate seat numbers.

time seq 20 | parallel -n0 "curl 'https://obcbzjejwjytmrlhiepnpklaobjjhpof.ihpapp.com/CreateReservation' -H 'content-type:
application/x-www-form-urlencoded' --data-raw 'libraryOpeningId=f2d102de-fe4d-4c21-9748-2219e095aa34&studentIdentifier=1234' --compressed"

It would have been great if on a Jobs definition I could define

instance Job ReservationJob where
    runMultipleJobsInParallel = False

[amitaibu]

Currently in order to prevent the multiple seats I'm using

perform ReservationJob { .. } = withTransaction do
    sqlExec "LOCK TABLE reservations" ()

So Maybe that's already doing the sequence I require.

[amitaibu]

Seems also master is still red

[mpscholten]

Another error :) fixed now

[amitaibu]

Thanks. It seems as if the queue is no longer processing. I've re-created DB and re-deployed on IHPcloud, but queue items are not processed (same on my local)

https://obcbzjejwjytmrlhiepnpklaobjjhpof.ihpapp.com/ShowLibraryOpening?libraryOpeningId=b7eba01b-88c9-4361-ae61-35e3b9475b88

On my local logs I see: NOTICE: trigger "did_insert_job_reservation_jobs" for relation "reservation_jobs" does not exist, skipping

[mpscholten]

Ah yes, the withTaskGroup is already stopped once it's callback function has returned. Therefore most tasks were directly killed after startup. Fixed this now on master

[amitaibu]

It works 🥳