We could monitor whether nodes are still alive.

[eduble]

Currently the booted flag of nodes is only updated in case of expected operations:

• set to False when the node is rebooted
• set to True when receiving a notification at the end of the its bootup procedure

Any unexpected issue will fail to have this flag updated. For example, logging to a node (walt node shell) and issuing reboot will make the node temporarily inaccessible, but the booted flag will remain to True.

Proposed update:

• the node could send keepalive notifications regularly.
• the server would consider the node is down if last notification is too old.