[tonnydan]

It is correct that one service application down in a services' cluster would not cause the service itself be offline, but the capability of processing requests will be impact. That is what I said which will impact on the SLA. And also, the reboot may not be predicted, which might be the fatal problem.

[usr42]

The reboot option is definitively not the right choice for all use cases but could be an option, depending on your constraints. At least it is still better then just not handling the expiration at all.

I think the follow-up post fits your needs better because no restart and so no downtime is needed. Hopefully I can finish it the next days.