Apparent Order

For the 6,000+ #AWS service API endpoints, awsipv6.neveragain.de shows which endpoints support #IPv6 by default, and which require additional client configuration. Important when considering an IPv6-only VPC. New: Daily updates and a list of recent changes.

@brankopetric00 I still think it’s a bad habit to use the `blame` synonym. Use `git annotate` instead. (This comes from CVS which also supports both commands)

git blame is called that because it accurately describes the emotional experience of running it.

@LeCodeNinja Hm!

Use system RAM as a GPU cache? gitlab.com/IsolatedOctopi…

@clovistb @myonlinetrust True, but let’s keep moving forward. I believe in a future without legacy IP.

@myonlinetrust @apparentorder IPv6 looks great but reality is still IPv4. most teams stay with dual-stack and NAT for now.

I recently asked: “What would you do if your VPC is running out of IPs?” A lot of answers were: “switch to IPv6” Lets be clear: IPv6 is NOT a quick fix. Switching to IPv6 is a full transformation 👇 1️⃣Upgrade your network Routers, firewalls, load balancers, VPNs must support IPv6 2️⃣Redesign IP addressing IPv6 is huge, but you still need structure. Plan CIDR blocks (/56, /64) 3️⃣Validate OS & systems Your servers, containers, and nodes must support IPv6 4️⃣Enable IPv6 in cloud VPC, subnets, ALB/NLB must support IPv6 5️⃣Update DNS Add AAAA records. No AAAA = no IPv6 traffic 6️⃣Rethink security No NAT in IPv6 - Everything becomes publicly reachable - Rewrite firewall & security rules 7️⃣Fix your applications Update configs, APIs, DB connections 8️⃣Choose a transition strategy Dual-stack (most common) NAT64 / DNS64 IPv6-only (rare) 9️⃣Upgrade observability Logs, metrics, tracing must support IPv6. Many tools still assume IPv4 1️⃣0️⃣Test everything Connectivity, latency, failover, DNS. Expect surprises

@myonlinetrust So it all hinges on delivery of the confirmation code e-mail, necessary to even start the sign-up.

EUSC seems to be very picky about e-mail providers it delivers to. It works for MS365 for example, but for at least two completely unrelated destinations, nothing arrives. I had a suspicion why, but that fell apart with the second failing destination. On the „real“ AWS partition it works fine though, so EUSC-specific configuration is at least one part of the puzzle.

Day 60 of not being able to sign up for the AWS European Sovereign Cloud. As far as I know, there are now two engineering teams and two support teams on the task of explaining failed e-mail delivery, with no meaningful updates for 43 days. 🤷

The mythical agent month

@steipete never put your middle leg on a moving sprocket

Hear me out...

@devopscdt @clovistb They needed to start gradual migration 20 years ago. At this point it's just ignoring the inevitable, fueled by vendor laziness. Even AWS has a hard time getting this right, it's embarrassing.

There are many reasons for this.Some of the below are as below. 1. Most of the existing apps(Work loads,load balancers,routers etc) run on IPV4. We need to reconfigure some or all of these to support IPV6 which would be a tedious task.Some applications may break as well. 2. Skill gap and we also need to update the security rules for IPV6. So the companies need to start gradual migration with proper testing.

In the cloud, IPv6 solves many networking challenges. So why are companies still holding on to IPv4?

@burkeholland Whatever is available for free on @kilocode 🙃

What’s your daily rn?

@zeeg That is what they call paying customers

The fact that datadog never named a project Good Boy is embarrassing tbqh

If you launch FreeBSD from the @awsmarketplace please be advised that I can now see your "Company name". I never used to be able to see this information, and have no intention of (ab)using it, but this is new. (And, of course, applies similarly with other Marketplace products.)

@tomwarren Great, now there is a market for old apple ids.

glad to see Apple using account age as verification for the UK’s Online Safety Act in iOS 26.4. This is how you do it 👍

@denkmit Seems to be surprisingly okay. Fingers crossed!

Seems like there’s been heavy rain and pretty extensive flooding at the Goiania circuit in the past few days. Situation doesn’t look great in some of the photos online but waiting to hear from people heading into the track today.

@kmcnam1 disk /huge_database has gone too long without being checked, check forced.

Perpetuum Slopile: AI use leads to more breakage -> more fixing needed -> AI to handle all that.

@edjgeek Both, I guess? Developers should (want to) understand what they're doing. Skills would be nice bonus material but not the key value.

Hey developers, question for you. Which do you find more helpful? Teaching you the patterns and best practices for serverless (and other technologies)? Or providing the right steering docs/skills/powers for your coding agents?

@myonlinetrust That’s kinda depressing because it shows AWS IPv6 doesn’t happen in whatever training data they used. Do you specify the endpoints explicitly instead of toggling the dualstack flag in the SDK?

@apparentorder Another place where Codex/ChatGPT keeps trying to “fix” things. “I see you’re using .api.aws as your endpoint. That actually doesn’t exist…”

AWS Cloudwatch feature launch: New endpoints for HTTP-based log ingestion. The endpoints are IPv4 only. aws.amazon.com/about-aws/what…

@myonlinetrust Oh. Yeah you’re right of course, can just use the dual-stack endpoint in both cases. Too early…

@apparentorder Hmm doesn’t .api.aws also work with IPv4? I know it does for most services.