#random - October 2025

Has anyone leveraged KEDA scaling for AmazonMQ for RabbitMQ queueLength metrics directly? For reasons I need to reach into the broker and get the queue's message count. I've tried to use the rabbitmq (which seems to not support https??) and metrics-api triggers. I think my issue is using https and not authing properly, but I'm not figuring out what I'm doing wrong. Doing curl with the same auth works, but KEDA gets a 401 back.

What do you think about AI bubble so far? Anyone seeing huge improvements(!)? Fully automated AI agents(!)? anyone? AI gurus? prompt
engineers(!)?
https://www.youtube.com/watch?v=YdpMVVUhYKM

In response to the October 20, 2025 AWS outage, thoughts about exhausting avenues to reliability on one's existing cloud provider first, about the difficulty of multi-region let alone multi-cloud, and about the importance of adopting the "enhanced region support" introduced in v6 of the Terraform AWS provider…

sqlxpert.github.io/2025/10/20/aws-outage-multi-region-before-multi-cloud.html

Do you really need your dev servers and databases at night and on weekends?

I'm excited to re-release 3 AWS cost-saving tools, now with Terraform support.

1. github.com/sqlxpert/lights-off-aws starts and stops EC2 instances plus RDS and Aurora databases, based on cron schedules in their tags.
2. github.com/sqlxpert/step-stay-stopped-aws-rds-aurora keeps infrequently-used RDS and Aurora databases stopped longer than 7 days.
3. github.com/sqlxpert/10-minute-aws-client-vpn demonstrates how Lights Off can save money by deleting and re-creating other resources that can't simply be started and stopped.
There's an any-region Terraform module for each tool. Lights Off and Step-Stay Stopped also have multi-account, multi-region Terraform modules. You can continue to deploy directly with CloudFormation, too.

I had planned to finish this release Monday, October 20, 2025, but we all know what happened that day. It turns out that a race condition and ties to a single region caused the AWS outage. On one hand it's the worst time to wrap up an AWS project, and on the other hand…

• I am proud that from the first release of its predecessor in 2017, Lights Off has worked independently in each region where it's deployed. AWS's much more complicated Instance Scheduler relies on a central DynamoDB database created by default in … you guessed it … us-east-1 !
• I am also proud that Step-Stay Stopped, which I wrote in 2025, avoids race conditions present in every public alternative I found.
Just as I demonstrated multi-region and multi-account deployment with CloudFormation in 2017, today I am demonstrating the enhanced region support introduced in the June, 2025 v6 Terraform AWS provider release. Though we can finally add region = to AWS resource definitions that we write ourselves, module publishers will have to catch up and pass this through.

One of the best defenses we have against inevitable outages is IaC templates that can easily be deployed in multiple regions and don't rely on central resources. My 3 examples are free and open-source, released under GPLv3 for people to learn from, improve on, and incorporate into other projects, so long as people reciprocate by sharing the what they build.

Have fun and keep building, in spite of alarmist reports that AWS (or cloud in general) is unreliable!

Hi all, the OSA conference is happening next week (online). Here are a handful of sessions you might find relevant, but highly recommend to see the full schedule on the website

• Emerging Architectures for Real-Time Observability at Scale — Peter Corless
• AI Your OTel — Mya Jaye
• Micromegas — Unified Observability for Video Games — Marc-Antoine Desroches
• ClickHouse® Chronicles: Real-World War Rooms with Human and AI Agents — Shivji Kumar Jha & Anurag Pandey
See the full schedule here: https://osacon.io/

🚀 Tetrix Overview — Giving AI Context Across Your Entire System
What if your AI could actually understand your code, infrastructure, and services — all at once?
With Tetrix, that’s now possible.
By connecting GitHub and AWS, Tetrix brings deep context awareness to your entire engineering stack — letting your AI reason across real repositories, environments, and cloud resources.
🎥 Watch the quick overview video here → https://www.youtube.com/watch?v=syoaxJla_EY

Hey everyone 👋

We’re starting the November matching round for our Dev Leaders Lunch Club. A curated community where senior engineering leaders get paired for monthly 1:1 lunches (virtual or in-person).

If you’d like to meet a new peer next month, make sure to join by Oct 30 to be part of the November pairings.

🎁 Bonus: new members this week will also be entered to win a free ticket to LeadDev Berlin (Nov 3–4) 🎟️

Sign up here 👉️ helloflea.com/lunchclub

Hey team! Quick read for today: Signadot’s CEO Arjun Iyer shared his thoughts on the “multicloud” narrative emerging after the recent AWS outage. Kubernetes seems like the obvious solution, but not hearing much discussion about it.
Curious to hear what you all think! https://thenewstack.io/the-great-aws-outage-the-11-billion-argument-for-kubernetes/

@Michael maybe an interesting post for you, working on terramaid. https://spidermonkey.dev/blog/2025/10/28/iongraph-web.html

Cloud GPU prices are moving faster than ever!

Our latest report tracks A100 and H100 GPU pricing across AWS, Azure, and GCP - and the swings are wild (up to 8x month to month 🤯).
As AI demand surges, the teams that stay flexible are the ones winning big:
• Moving across regions
• Shifting between clouds and neoclouds
• Using automation to always grab the most cost-efficient GPU
It’s not just about saving money - it’s about building resilience in a market that changes by the week.

👉️ Read the full report here

Hey Everyone ,

I’m excited to share that our product, Tetrix, is officially live! 🎉

Tetrix connects your code, infrastructure, and operations directly to your AI — enabling it to reason across your entire software system. I genuinely think it could add a lot of value to what you’re building.

It’s free to try, and you can get started in just minutes by connecting your GitHub and AWS.
👉️ Try Tetrix now-https://deskree.com/

I’d really appreciate any feedback — good or bad.