M365 Outage: Automation Bug Pulled Too Many IP Routes
Microsoft traced the July 23 M365 outage to a bug in its automated maintenance system that removed IP routes from more network devices than intended, taking down Azure and M365.
Loop Microsoft’s automated network maintenance request system sent a change to network devices on July 23 that removed IP routes from more endpoints than it was scoped to touch. The result was immediate and wide: Azure infrastructure and Microsoft 365 services — Teams, SharePoint Online, Excel Online, and the M365 Admin Center — went unreachable until the routes were restored. BleepingComputer has Microsoft’s post-incident statement.
Microsoft attributes the failure to a bug in the maintenance request system itself, not to an operator who entered the wrong scope. That distinction matters. A human misconfiguring one change produces one incident. A bug in the automation layer means the same over-broad scope can run every time that maintenance job fires until the underlying logic is corrected.
The initial outage report noted that the M365 Admin Center went down alongside the user-facing services — Teams and SharePoint were joined by the tooling administrators would use to investigate and remediate. The management plane and the data plane share infrastructure. A routing change that takes one down takes the other down at the same time.
A separate Exchange Online mailbox quarantine error ran concurrently. Microsoft has not connected the two incidents. Both are classified as independent platform failures.
Check now: Confirm both incidents are resolved in your tenant at admin.microsoft.com → Health → Service health. The post-incident review — covering the scope of the routing change and what was corrected in the maintenance system — will appear there. That document is worth reading before the next automated maintenance window runs.
Found this useful? Share it.
