Hey Neight,
The supervisor tries to conserve network traffic as much as possible by sending messages only when it is necessary. The drawback to this is if a machine has been shutdown, the supervisor will not notice it for a period of time. This however changes once the supervisor tries to send a job to the host. It is then marked as unavailable. The "active" state doesn't take any qube! licenses, so typically it's ok to just let the supervisor "think" the hosts are still active. You can shorten the period of detection by setting up the workers and the supervisor to use a heartbeat check on each other.
To enable it, turn on the "Heartbeat Monitor" in your Config gui under Supervisor Settings->Advanced Settings
If you wish to tweak the timeout settings, then you'll need to edit the qb.conf file directly. The default timeout before the supervisor checks on a host is 12 to 14 minutes. You can tweak this by changing the following setting in this file:
Windows C:\winnt\qb.conf or C:\windows\qb.conf
Linix/OSX /etc/qb.conf
supervisor_heartbeat_timeout = <number of seconds for timeout>
You will need to restart the supervisor once these settings have been changed.
Thanks,
Anthony
