The context · June 2025
Cloudflare’s June 2025 postmortem described an outage affecting Workers KV and dependent services, triggered by a failure in underlying storage infrastructure involving a third-party provider. [1]
The shared dependency matters
A system can use several named services while still depending on one underlying component. When that component fails, apparently separate functions may fail together. Looking only at the application’s immediate providers gives an incomplete picture of its availability.
Start with the customer’s task
Trace a critical journey such as signing in, finding an order or submitting an enquiry. Identify which dependencies must work at each step. Decide where a safe read-only mode is possible and where the application must stop. Authentication failures should not be handled by casually bypassing access controls.
Turn the postmortem into a rehearsal
The lesson for a smaller team is practical: rehearse losing a dependency and inspect the experience a customer receives. A clear failure message and a tested recovery process can be more useful than an elaborate fallback nobody has operated. This case concerns Cloudflare’s published incident, not an outage or engagement at Sansa.
Source & context
Cloudflare · Service outage postmortem, 12 June 2025This retrospective was written for the archive in September 2026. The linked primary source documents the announcement or event; the practical interpretation and proposed approach are Sansa’s editorial perspective. Public examples do not imply a client relationship. Product capabilities and guidance may have changed since the period discussed.
Another perspective · June 2025
A dependency-loss rehearsal for a customer portal
Continue readingWorking through a similar question?
Talk it through with Sansa