Hi, We are implementing a feature which relies on the fact that the executor for periodic tasks is shut down on drain which is not happening now (o.a.c.concurrent.ScheduledExecutors.scheduledTasks).
I want to ask if there is some reason we are not shutting down all executors in StorageService#drain but non-periodic ones. There are executors for scheduledFastTasks, scheduledTasks, nonPeriodicTasks and optionalTasks but we are shutting down only the executor for nonPeriodicTasks. Does somebody know why? Is there some very specific reason behind that or we just forgot? Are there any (to me yet unknown) consequences of shutting them down? My wip patch (1) with build (2) seems to be just fine when I shut them down. We would like to see this in 4.1. Jira issue in (3) (1) https://github.com/apache/cassandra/pull/1529 (2) https://app.circleci.com/pipelines/github/instaclustr/cassandra/882/workflows/5eba531c-433c-4112-abcd-dd9622188170 (3) https://issues.apache.org/jira/browse/CASSANDRA-17493
