|Contact Person:||Jared Dunbar|
|Last Update:||July 2017|
|VM Host (Physical):||hydra|
|Hostname:||dns2.cslabs.clarkson.edu, dns2.cosi.clarkson.edu, atlas.cslabs.clarkson.edu, atlas.cosi.clarkson.edu|
|Operating system:||Not Debian, likely Arch or Ubuntu|
|Development Status:||Waiting for Fall 2017 to start development|
Atlas will be a DNS/DHCP server which will act as a failover for essential network infrastructure, and assist Talos as not being the single point of failure.
This part will be easy, since it will be filling the place of dns2.cslabs.clarkson.edu, which previously existed and still is in the nameserver delegation rules.
It will be synchronized with the git repo for Talos automatically every 5 minutes.
This will be the hard part. We don't want to give out double leases (or do it as little as possible), so we will use the existing Talos leasing system, and when Talos goes down, this service will realize when the leases expire, and start giving out new leases until Talos hands out leases again, at which point it will let all of the leases expire. Since Talos and Atlas will both have 5 minute lifespans on their DHCP leases, this could cause a few minutes of overlap when Talos goes up or down, but not too much of an issue.
If someone wants to, they can set this up. Technically it's not too hard to set it up so that services can contact one or the other, but synchronization may be difficult.