SRE Weekly Issue #431

This is a really thorny one. As individual subprocesses started infinitely looping, their system shifted load to other datacenters, masking the problem. A coinciding failure in the load shifting system made things even more interesting.

  Lloyd Wallis, Julien Desgats, and Manish Arora — Cloudflare

A great discussion of where dashboards fall short and what we should look for instead.

  Adam Kinniburgh — SquaredUp

Read how we have significantly improved the ability of our monolith to correctly and fully process pushes from our users.

  Will Haltom — GitHub

Timing things to happen at specific intervals is yet another way that we collectively find out that dealing with time is a hard problem.

This article illustrates the subtle but important pitfalls in trying to create a system that does something on a strict interval.

  rachelbythebay

This article reads more like a case study. The author gave a prompt to three different LLMs and actually tested the Terraform config it produced.

  Mike Vanbuskirk — Terrateam

When your pub/sub system can have a million subscribers, even something mundane as notifying about subscriber counts requires careful thought.

  Ashmeet Singh — Pusher

To me, this concept comes up over and over in SRE, and it’s a core part of SLOs.

  Juraj Masar — BetterStack

In this blog post, we’ll dive deep into the technical aspects of feature flags and feature management, exploring how they can be leveraged by SREs to enable progressive delivery, improve system resilience, and optimize the user experience.

  Hope Lynch — CloudBees

This week’s Mentour Pilot video covers an accident that involved an inaccurate flight simulator. I wasn’t familiar with the term “negative training” before, but now I’m going to be keeping an eye out for it in the systems I manage!

  Mentour Pilot