There are cases when the scheduler receives a SIGKILL between the zrem and zadd from enqueue_job(). This would, in turn, lead to occasional unscheduling of the repeatable jobs.
The solution I propose is to put everything in an atomic transaction and thus assure atomicity.
There are cases when the scheduler receives a SIGKILL between the
zrem
andzadd
fromenqueue_job()
. This would, in turn, lead to occasional unscheduling of the repeatable jobs.The solution I propose is to put everything in an atomic transaction and thus assure atomicity.