In the current cron adapter implementation, the adapter subscribes to all cron flows. If two instances would run, both would subscribe and trigger them.
To solve this problem, the cron adapter should be a nats group that subscribes to the topic.
Every adapter will send a table every view minutes to mark itself as alive, if one does not send the table he will be declared as death and the table will be split to all left adapter to get the flows of the dead cron instance