Aidan W Steele (@__steele) 's Twitter Profile
Aidan W Steele

@__steele

I try to tweet novel things about AWS.“Shit-poster extraordinaire” according to @LastWeekInAWS. He/him. AWS Serverless Hero

ID: 281819105

linkhttps://awsteele.com calendar_today14-04-2011 01:23:40

12,12K Tweet

8,8K Followers

2,2K Following

Aidan W Steele (@__steele) 's Twitter Profile Photo

It would be nice if Terraform automatically (or at least had the option to) tag every resource with its "resource address" My workflow often starts with finding a resource in AWS and trying to figure out where it came from, sometimes with deeply-nested modules. This would help!

Aidan W Steele (@__steele) 's Twitter Profile Photo

AWS CodeBuild seems to have just got a *tonne* of new IAM policy condition variables, and I am delighted to see what this might mean for other AWS services. It looks like just about every request parameter can now be used in policy evaluation 🤩 (h/t to Jon Rault for noticing)

AWS CodeBuild seems to have just got a *tonne* of new IAM policy condition variables, and I am delighted to see what this might mean for other AWS services.

It looks like just about every request parameter can now be used in policy evaluation 🤩

(h/t to <a href="/jo_n_go/">Jon Rault</a> for noticing)
Aidan W Steele (@__steele) 's Twitter Profile Photo

What's the state of the art for deploying an ECS service from a CI/CD pipeline? E.g. I've just built and pushed an image, how do I best say "ECS, use this new image tag and return an exit code depending on whether you successfully rolled it out" (with event/log streaming, etc)

Aidan W Steele (@__steele) 's Twitter Profile Photo

Why does AWS Amplify not use CodeConnections? The latter is a nice way to set up integration with GitHub once and share it across a whole org - except for Amplify 🫠 Is it due to Amplify having extra functionality, maybe?

Aidan W Steele (@__steele) 's Twitter Profile Photo

Why do AWS ALB health checks expect gRPC status code 12 (UNIMPLEMENTED) instead of 0 (OK) by default? I just spent an embarrassingly long time trying to figure out why the ALB considered my service unhealthy. That'll serve me right for trying new things and skimming the docs.

Aidan W Steele (@__steele) 's Twitter Profile Photo

Has anyone written an article “so you’re a SaaS looking to integrate with customers’ AWS/GCP/Azure accounts. Here are the do’s and don’ts”. I feel like that would be useful to share with vendors who make suboptimal choices.

Aidan W Steele (@__steele) 's Twitter Profile Photo

My favourite thing about Pocket was that I had a browser extension on desktop and mobile that allowed me to save articles in 1-2 taps and that I could read them on my Kobo e-reader later. How can I reproduce this workflow after Pocket shuts down? New device is fine if it works!

Aidan W Steele (@__steele) 's Twitter Profile Photo

This isn't true, is it? I feel like AWS might have copy-pasted the verbiage from SCPs and done a search and replace to "RCP" (especially because the link goes to a URL named "scp-inheritance"!)

This isn't true, is it? I feel like AWS might have copy-pasted the verbiage from SCPs and done a search and replace to "RCP" (especially because the link goes to a URL named "scp-inheritance"!)
Aidan W Steele (@__steele) 's Twitter Profile Photo

How do you make Iceberg play nicely with S3 access points? Iceberg metadata files contain fully qualified S3 URIs, but in a data lake you probably want different consumers using different access points. I see Spark has mapping logic, but what about other Iceberg clients?

Aidan W Steele (@__steele) 's Twitter Profile Photo

I'm using GitHub Actions. I have a protected environment: prod. I want to deploy to production after a successful deploy to staging. Not every commit will go to production. Newer workflow runs should cancel pending ones. Pending shouldn't appear in deploy history. Doable?

Aidan W Steele (@__steele) 's Twitter Profile Photo

This is so good. Now I just want GHA to implement OTEP 258 and I’ll be the happiest dev in town. github.com/open-telemetry…

Apparent Order (@apparentorder) 's Twitter Profile Photo

When AWS announced their IPv4 tax, >90% of service endpoints did not support IPv6 at all. This number is now down to 33%. To celebrate, I gave awsipv6.neveragain.de a makeover – it's now dynamic, so it won't shove several megabytes of HTML into your face anymore. Also, to keep