Infrastructure October 19, 2026 bearish ⇧ 2433 pts across 3 threads

Shared infrastructure keeps surprising people

Let's Encrypt is adding two more remote perspectives for domain validation, and the top comment points out that in August an attacker still got a certificate issued despite multi-perspective validation, after a rival network hijacked a block of Hetzner-hosted infrastructure for 33 hours. A thread on Docker pulls failing in Spain explains that La Liga blocks Cloudflare, including R2, during football matches, and Docker Hub is collateral damage. In the AWS billing thread, hobby accounts that normally cost under $1 a month saw estimates of hundreds of millions and even billions during a $1.7 billion estimated-billing error.

The pattern: a handful of providers sit under everyone, and their failures arrive as legal blocks, routing attacks, or bad numbers, not just outages. Commenters keep reaching for the same fixes, a VPN in the pipeline or running outside Spain, which tells you there is no clean answer.

The counterpoint is that these providers fixed things people used to find hard. One commenter noted that HTTPS was once difficult and expensive before Let's Encrypt.


So what?

Build for the case where your CDN, registry, or cloud console misbehaves. Mirror critical artifacts, set hard budget guards that don't rely only on the provider's estimates, and don't assume your CI can always pull from one registry.

Read these