Open MattMills opened 6 years ago
@dneuman64 @jhg03a - is this still required with the implementation of https://github.com/apache/trafficcontrol/blob/master/blueprints/multi-interface-servers.md
Server interfaces doesn't address this request
I agree with Jonathan, the intention is to provide a mechanism by which TC itself can be "aware" of expected periods of downtime and route around them with no other operator intervention.
Server interfaces doesn't address this request
agreed. just wondered if a server having multiple interfaces reduced the need to even admin down servers for maintenance purposes. i guess not.
This ticket is predicated on queue and snap being safe for automation to perform.
If it was possible for an Admin user to add a "maintenance" event to a server (Start time, stop time , required reason, username), that data could then be used so that when the server runs ORT within that time period the actual server status (if status == reported) is overridden with ADMIN_DOWN.
This would enable ops people to pre-program the network maintenance and other stuff we need to work around, as well as give us a fantastic way (much more visible) to keep track of it for troubleshooting purposes.
In the self service world, this could even be taken a step further of just giving the permission to setup maintenance periods to the networking team directly.