On December 3rd, SDM released a server build that added a new index to the table tracking latency between nodes. While the build succeeded on the EU and UK Control Planes, it failed on the US Control Plane.

Retrying the build on the US Control Plane left the index in an invalid state, which went undetected by our migration tools.

The following day, December 4th, SDM released a server build that relied on the new index. Without the index, node-to-node latency was no longer stored, and approximately three minutes later, the routing system stopped attempting multi-hop routes. This issue was limited to the US Control Plane, as the index was successfully created in the EU and UK Control Planes.

StrongDM has already taken steps with internal processes to ensure that this kind of issue does not repeat in the future. Thank you for your patience and understanding.