Tomorrow we ride at dawn
Tomorrow we ride at dawn
Tomorrow's training agenda:
[I'm really liking these CodeWindows BTW - just little peeks at your real files so the 🪿 don't decide to apply their own brand of middleware on the way through when they're writing your notes]
#engineering #platformengineering #teachingthought #teachingcommunity
Y'all built a CLI Jenkins console log tailer too right?
So much nicer without all the noise they throw in (does anyone care that [pipeline] did a thing again)
And colours
This was a while ago though... I think I can do better
The 👶🏻🪿 (a wee GPT we pull in for transcript summaries of our training session) has spoken:
Today's training summary:
#OverloadProtection is one of the most overlooked practices in #PlatformEngineering.
The result? A fragile patchwork of inconsistent fixes across teams.
Centralized rate limits, quotas, adaptive controls, and real-time visibility give services predictable ways to absorb traffic spikes, reduce reliability debt, and prevent cascading failures across systems.
Learn more on #InfoQ 👉 https://bit.ly/48y5Bak
Did you miss Charlotte, Matt, and me talking about #GitOps #PlatformEngineering and #Compliance at SHIPPED25?
You can watch on demand here:

And then another one, a single pipeline orchestrating all of the stages, two AMI builds for the two tiers, rolling, controlled deployments of the front and back, full automation of the HAProxys' views of their target backends - all packaged up itself in a single file that describes both preproduction and production releases. The ultimate in testing your release automation, the configuration itself is DRY between environments, a slide rule window over the context that fills the voids the only difference.
[my other baby, a baby eagle called Peregrine - my developer platform and I'm quite proud of her]
The fact that they had all of this sitting right in front of them and decided that the thing to do was call me... might be why all the rest of that #wanderland nonsense is normally what I'm posting 🪿🧸🌱🎩💜
#devops #platformengineering #ileftanoutageforthis #witheverythingtheyneeded #neuroawesome #nomoreburnout
Oh yeah, I left an outage call to come work on this.
I mean, I left a technical DevOps manager, a DevOps engineer, a Lead Software Engineer and a Principal Technical Architect with a screen shot of the launch template and a big arrow pointing at the version to roll back to.
I don't even own the system, I just happen to like the guy that did and I helped him setup the automation. It's slick, two tier Shibboleth + HAProxy with redundancy on the front end and active/passive failover on the Shibboleth tier.
Each component has it's own deployment configuration. A single YAML that describes the process from little baby Packer template to the fully formed AMI that's cloned into the production environment. All the bits and bobs, URLs, pointers to infrastructure templates. Everything you'd need to know to deploy the thing (since it's how the thing is deployed... where else would it be?)
#engineering #platformengineering #devops #ileftanoutageforthis #witheverythingtheyneeded
We're still doing that thing where we feed the transcripts of our training sessions back into our knowledge graph to compare to the projected agenda for the day?
Make sure we cover everything before we plan the next day's agenda?
Feed it through your in house AI API to extract our the salient details for free? (to you anyways).
Just cuz you setup all their pipelines on your developer platform and they showed you how to get an auth token?
Just cuz you keep telling them that they can cache their huggingface models on shared storage and not keep baking them into the image they're deploying to Fargate that now takes an hour to bake and almost as long to start up when they need to scale?
(TMI?)
We're still doing that right cuz it's neat.
[Takes longer to fix the permissions on the transcript in sharepoint than it does to perform the analysis and feedback loop]