Microsoft’s routine device maintenance on July 23rd set off a regional Azure outage that cut traffic in West US and affected 27 services. What was meant to be an impact-less maintenance window instead removed network routes from more devices than intended, and the disruption lasted almost five hours before full recovery.
Multiple Azure services began to detect and correlate service degradation a minute after maintenance began at 14:44 UTC, or 07:44 AM Pacific Time. That early warning mattered because the incident spread quickly across Microsoft’s West US region, where traffic entering and exiting the region was affected as routing began to churn inside the company’s Wide-Area Network.
Microsoft later said a bug in the request conversion system incorrectly marked additional devices as part of the maintenance event. That mistake caused a set of IP routes to be removed between its datacenter and wide-area network, which is why traffic was disrupted even though the work was supposed to be isolated and low-risk. The company also identified recent fiber maintenance activity between 16:00 UTC and 17:45 UTC while tracing the routing behavior.
The failure point was not a broken service in isolation but the maintenance machinery itself. Microsoft said the process converted requests into system-readable instructions and was supposed to keep at least one of two redundant paths healthy, yet the bug pushed the route removal beyond the intended devices and took out 27 services before the company began rolling back changes at 17:45 UTC.
By 18:26 UTC, Microsoft said its WAN was back to its best, and all impacted services had fully recovered by 19:41 UTC. The sequence leaves one question that matters to anyone running on Azure: which specific services were among the 27 affected, and how far beyond West US the fallout would have spread if the rollback had started later.
