Summary: there is a routing issue.  We understand the problem.  There are a couple of solutions.  We are working to get things back up as soon as possible.

Update @UTC 1032: IPs are again pinging.  If you are experiencing any issues, please let us know (via a ticket?).  We will do a debrief with the data center staff tomorrow.  And follow up here when we hear back.

Update @UTC 0959: The cogent engineers are migrating over fibre connections to the new gear.

Update @UTC 0918: The new cogent gear is being powered on.  If things go well we should expect things to work 'shortly'.

Update @UTC 0809: The data center report:

Advertisement takes up to twenty four hours, while this needs to happen and while we will make sure this is done correctly, this will not provide immediate relief. We have contacted our senior networking staff to see if anything else can be done, and Cogent is proceeding well in their work, the new hardware is in place and they are feeding the DC power cables at this time.

Update @0743 UTC: Some detail: Back in 2012 the data center were advertising our 199.231.76.0/22 range through cogent and another provider IPTransit.  They subsequently replaced IPTransit with a different provider.  But did not update our IP range with that provider, leaving that range only advertised with Cogent.  So now that cogent is down there are routing issues.  We will be advertising another provider shortly for those IPs, but that may take a few hours to take effect.  Cogent have scheduled a maintenance window up until 0600 CST (1200UTC).  We would expect any issue to be resolved by then but hopefully a lot sooner.

Update @0734UTC: The data center report Cogent were doing hardware maintenance.  Which is normal/typical.  That work should _not_ have resulted in the level of issues we are seeing.  It appears to only affect IPs in the NetRange: 199.231.76.0 - 199.231.79.255 (199.231.76.0/22).  We are checking what options we have.  e.g. using different routes.  e.g. getting an ETA from cogent.

Update @0652UTC: Cogent is currently performing maintenance on there router in the data center. During the maintenance window you will notice subpar routing and higher than normal latency. We will update the notice once we have confirmation that cogent has completed there maintenance.

Some servers in our Dallas data center appear to be inaccessible from different locations. We're investigating and will update this notice as soon as we have further information.