This message was deleted.

Controversial opinion - don't reboot, fix your applications instead 😉

At this point I am satisfied with a weekly reboot 🙂

That was slightly tongue-in-cheek, I know it is hard to get away from reboots. Only managed it in one deployment

Weekly reboot here to update MCS image. Custom script to ensure a minimum capacity, maintenance mode > drain on inactivity > restart

Extra capacity with rolling reboots.

With sufficient capacity, you can query the VDA uptime, when 24h of service/uptime is reached, enable maintenance mode, drain on inactivity, and finally restart and put the VDA back in service. This is how I would do it.

VDA split into several groups (through machine catalogs or tags), custom script to drain and reboot

What about "Restart after drain"? When the restart time is reached, machines are put into the drain state and then restarted when all sessions are logged off. Works for me.

You could go in these ways, the customer just needs that the connected user does not have to reconnect during the reboot of the vda.

True. In my case Idle and Disconnect times are rather strict. Both set to two hours and so normally the VMs can drain all sessions without facing this issue.