On February 26th at 15:46 UTC, we started routine maintenance in our London data center. At 15:47 UTC, customers began to experience issues starting new Voice API and WebRTC calls, and the inability to control in-progress calls. Our engineering team were alerted to the issue at 15:48 UTC and restored all services at 15:53 UTC. This issue impacted Voice API and WebRTC calls only - Messaging, SIP and WebSocket traffic did not experience any impact.
During routine maintenance, a bug in our deployment software caused a segment of voice services in our London data center to enter a failed state.
Upgrade deployment software.
Improve routine maintenance testing and scheduling processes.