networking

I recently learned that AWS internet gateway does a one-to-one destination NAT for public IPs inside a VPC, this includes public load balancers and elastic IPs (EIPs) on EC2 instances. So for any inbound IP filtering rules I need get the private IP addresses of the resources (if filtering by destination IP).

The closest documentation I could find of this behaviour talks about this in regards to EIPs:

traffic that's destined for the public IPv4 address or Elastic IP address of your instance has its destination address translated into the instance's private IPv4 address before the traffic is delivered to the VPC.

Seems logical in hindsight but led to some fun troubleshooting.

Related post on Mastodon.

I'm a fan of discovering weird places to host web servers from and a disposable vape pen microcontroller might be the weirdest. Bogdan Ionescu setup their server on a PUYA PY32F002B with 24KiB of flash storage and 3KiB of RAM. They used slattach to convert the serial connection to the microcontroller into a network interface and uIP for the TCP/IP stack. A lot to unpack in the full article.

Discovered via The Morning News.

Xe worries an incident report after a flapping IPv4 BGP session with their hosting provider affected the availability of some services over IPv4.

1. An important takeaway:

   The Uptime Kuma instance that caught this incident runs on an IPv4-only network. If it was dual stack, this would not have been caught as quickly.

My machine running status checks via Uptime Kuma is dual stack. This is something I need to look into as well.

TIL you can use an Apple TV as a tailscale exit node which according to the Mastodon thread also works when the app is running in the background. Cool use case. Being able to setup GitHub actions to use the tailscale exit node is a bonus.

Coincidentally, I also shared a post earlier today from Ruben Schade about his Apple TV.