Currently when we return a 429 to an integration action, the action will retry (or at least is supposed to @basdebruin) at 60 seconds from when it last tried. I have a separate request open for honoring retry-after headers.
Scenario: User has changed status on 500 items, triggering an integration. Monday queues these 500 operations and begins processing (I think its about 20 initially). If we return a 429 with a retry-after then the entire queue should pause. Any actions with a 429 should be considered outstanding still and keep the max outstanding requests for the queue
Currently, the action will retry after 60 seconds (from its initial invocation). But while it waits to retry, the queue continues to process - the remaining 489 actions in the queue all get sent to our app back ends, and all receive a 429.
Monday should pause the queue processing for that integration if an integration has received a 429 until the 429 is resolved. This will reduce load on app developer back ends, on monday code, on the monday infrastructure, etc.
Further, with the 429 we should have a way to return a “pause all app integrations” request. This would be if API complexity is hit and we can’t process further actions - regardless of which integration. This would cause all new actions to automatically queue and wait for the app until the retry-after. Now there is no further request load on servers until that time.
Significantly, this will reduce energy consumption. Making our apps greener should be a goal, right? It is supposed to be - ESG at monday.com