Blog
StackWitness blog
Honest notes about measuring dependencies from the outside. We ping hosts for reachability and read vendor self-reports. We do not invent green, and we do not issue compliance verdicts.
-
When one zone loses power, the other two can fill up
On 20 October 2026 a power cut took Google Cloud zone europe-west2-b offline at 06:00 UTC. Failover then filled the other London zones. How to say that while regional APIs are still returning 503.
-
When the Azure gateway list kept moving
Microsoft's Azure gateway incident from 20:30 UTC on 30 September 2026 to 02:15 UTC on 1 October started as ExpressRoute in UK South. By the end the public list held five products, a corrected region note, and five regions still changing at 01:36.
-
When the replicas detach and the host still answers
GitHub's 23 September 2026 incident started as detached database replicas: organization creation, the API, and Projects degraded while Git Operations stayed off the affected list. How to keep those clocks separate when the component board goes quiet before the incident closes.
-
When the edge returns 1016, the origin is still down
Microsoft Copilot's website returned Cloudflare Error 1016 from 22:25 UTC on 9 September 2026 until 00:05 UTC on 10 September. How to read an edge error page without declaring the CDN down, or the product up.
-
When the cron is late and the host still answers
Cloudflare's 9 September 2026 Workers Cron Triggers incident left HTTP paths answering while scheduled jobs could miss or delay. How to separate a 200 from proof the job ran.
-
Three AI outages on one morning are still three incidents
On 3 September 2026, ChatGPT, Claude, and Grok failed in overlapping windows. How to write the customer update without inventing one shared root, and without calling the whole industry down.
-
When one zone is down and the region still looks up
Google Cloud's 1 September 2026 incident hit a portion of us-central1-b while the rest of Iowa kept answering. How to name a zone failure without averaging it into a calm region badge.
-
When the control plane is down and the data plane is up
DigitalOcean's late-August API and control-panel incident left running Droplets up while creates and dashboard access failed. How to separate control-plane pain from host reachability without inventing calm green.
-
What to put on a customer status page during a vendor incident
When a dependency is degraded and your product feels it, customers need measurement language, not a calm green badge. A practical template from the August 2026 GitHub outage.
-
Partial outages: why one green average lies
When one product is red and the rest look fine, a single operational badge hides the real story. How to read component status, region disagreement, and host reachability without inventing calm green.
-
Is it you or your vendor? A 3am playbook
When checkout fails at 3am, stop guessing. A practical playbook for separating your stack from a dependency using independent reachability and vendor status pages.
-
Why vendor status pages cannot be trusted alone
Vendor status pages are useful self-reports, not independent measurement. Why teams need an outside reachability check before telling customers everything is fine.
-
How we measure: reachability vs self-report
How StackWitness measures dependency hosts: multi-region reachability probes plus vendor status feeds. What reachable means, and what we never claim.