A strange issue is found that the Alteryx Gallery schedule suddenly becomes disabled. After checking with support, it looks like the mechanism underlying is using a concept of queueID. Whenever there is a network issue between controller and worker, the queueID get scratched, the job gets corrupted. This means it cannot calculate the next run and thus, it will disable the schedule. When the worker node gets back, it cannot "try again".
From usage perspective, if network is having issue, it is assumed that the connection gets back, it would "re-try" and resume.
Hope Alteryx team can help consider a fix on this technical issue.