Service Outage: October 5, 2026

What happened, what we did, and what comes next

Date October 5, 2026

How long - About 68 minutes (3:05 PM to 4:13 PM Mountain Time)
Who was affected - All users
Status today - Service is fully restored. One behind-the-scenes task is still in progress (see below).

What happened

On the afternoon of October 5, our services went offline at our main data center. The PDU (Power Distribution Unit) in our cabinet overloaded and tripped one of the breakers. When the load was transitioned to the backup power leg, the same breaker on that PDU also tripped. This caused a loss of power for our network equipment. We moved our services to our backup data center, and they were back online about an hour later.

Why it happened

The electrical system in the data center was carrying an uneven load, with too much demand on one circuit. That circuit overloaded and shut off, which cut power to the equipment that connects our systems to the internet. The facility has two separate power paths allowing the backup to take over if the primary fails. In this case, both were overloaded, one after the other. Resulting in the redundant power path not being able to prevent the outage.

How we fixed it

Our team switched services over to the backup data center, which restored service at 4:13 PM. Technicians then went to the main data center, checked that our data was safe, evened out the electrical load, and turned the power back on. By 5:05 PM our servers in the main data center were running again.

What is still in progress

Our services are running using the database servers in the backup data center. Our main data center’s databases still need to be reconnected and resynchronized. While that is being done, we still have a pair of servers running to provide high availability. We plan to finish this during a scheduled maintenance window, which is expected to cause little or no downtime.

What we are doing to prevent this

• Reviewing the electrical load at both of our data centers and evening it out.

• Evaluating real-time alerting to warn us when power usage gets too high, before anything shuts off.

Timeline

3:05 PM - Connection to the main data center was lost.
3:57 PM - Working with an onsite data center technician, we determined the network equipment had no power and could not be powered on. We began the failover to the backup data center.
4:13 PM - Services were back online from the backup data center.
4:50 PM - Our technicians reached the main data center, found the cause, and restored power.
5:05 PM - Servers at the main data center were running again.