For some additional clarification, a cron task is not necessarily a task queue task. As Alejandro said, it can merely an HTTP request. That means it can be of the following:
1) a request sent to a standard autoscaling instance (F* instances), be subject to a 60 second deadline; or
2) a request sent to a handler that triggers a task queue on a standard autoscaling instance, subject to a 10 minute deadline; or
3) a request sent to a basic or manually scaled instance (B* instances), subject to a 24 hour deadline.
When using cron, be aware that failures will only be retried at the next specified execution interval, except in the case of an HTTP 503 response from the target handler. In that case, the request will retry indefinitely, as short-term quota exceptions return HTTP 503 and are typically safe to retry.
Cheers,
Josh