Scale down can happen when worker deletes the job before processing

[practo]

Scale down can happen when worker deletes the job before processing

[alok87]

Yes but deleting the job then doing the processing is a bad design. Job should be deleted after the processing is done. But true! WPA currently does not support this use case. Won't count this as bug so removing the tag.

[anuragagarwal561994]

It depends on whether it a good / bad design based on use case we are trying to tackle, can you have an article to support your statement.

Arguments in support where this can be useful is when there are long running jobs and you don't want the job to be cancelled and repeated again once the timeout expires, one wants to take control at application level of how a job should be scheduled next if it fails once.

[alok87]

This usecase is not supported at present.

[alok87]

Closing this, since this behaviour is incompatible with any scaling logic.