We manage several customer-owned tenants in West US 2 and West US 3, and Azure Storage has started failing intermittently with permission-related errors. The failures have become more frequent over the past hour and are affecting multiple unrelated customers, so a single service principal or resource-group configuration problem seems unlikely.
Neither the public Azure Status page nor the tenant-specific Service Health view is showing an incident. We already have two Sev A support cases whose response SLAs have been breached, leaving the overnight team repeatedly checking for updates so the day shift can continue troubleshooting.
If production is failing across multiple tenants while both health pages remain green, how should this be escalated and what information is most useful for proving that it is an Azure-side issue? Also, what is the practical purpose of the public status page if incidents like this do not appear there?
3 Answers
The most useful escalation package is a short timeline showing when failures began, which regions and tenants are affected, the specific Storage accounts and operations involved, representative request IDs, and whether retries succeed. Avoid describing it only as a generic permissions problem; include the raw response details so the engineering team can distinguish authorization failures from a control-plane or backend fault.
The public Azure Status page is not intended to be a real-time dashboard for every service problem. It generally lists incidents that meet Microsoft’s criteria for broad public impact. A green status does not guarantee that every region, tenant, or operation is healthy. Service Health is meant to provide more targeted notices, but it can still lag behind an emerging or narrowly scoped incident, so direct escalation with detailed telemetry is still necessary.
This sounds more like a backend Storage issue than a shared credential problem. Collect the exact Storage error codes and the x-ms-request-id or correlation ID from several failures across different tenants, subscriptions, and regions. Include timestamps, affected operations, and the scope of the failures in every support case. Showing the same pattern across unrelated customers is strong evidence that the issue is on the platform side and should help Support correlate the incidents.
The customers own the support plans, but they are all on premium enterprise offerings. We are passing the request IDs and cross-tenant impact to them so their cases can be correlated.