💾 Archived View for jsreed5.org › log › 2023 › 202305 › 20230506-extended-downtime.gmi captured on 2023-06-16 at 16:27:27. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-05-24)
-=-=-=-=-=-=-
---
Rob's Gemini Capsule was offline for a full week, from Friday to Friday. The downtime also caused my contact e-mail address to fail, resulting in bounced messages.
We moved to a house in a different city over the weekend, which necessitated us signing up with a new ISP. Our previous ISP gave us a static public IP address by default that resolved directly to our home router, so setting up self-hosting services was trivial. Our new ISP requires a business-level account to receive a static IP address, and they charge an extra fee to bridge that public IP directly to our home router.
The extended downtime came from confusion regarding what I needed from their infrastructure. We started with a residential plan with no extra features, and I tried to use a dynamic DNS to resolve the jsreed5.org hostname to our non-static IP address, but after configuring port 1965 in my home network I was still getting timeout errors. Upon calling the ISP, I was told they needed to forward ports in their own equipment, which they did. However, the capsule still did not resolve.
Two more calls secured me a "public" IP address (according to their own terms), and then a straight upgrade to a business account with a static IP address. After those three calls, though, the problem was still not solved. Finally, a network engineer informed me that my static IP address was still routing through their own internal VPN infrastructure, inside which my WAN interface was connecting to the VPN gateway to access the Internet. For an extra fee, they could bridge the public IP directly to my WAN interface, logically bypassing their systems completely. I agreed to do it, and the problem was finally fixed.
The ordeal puts into perspective how difficult it is to self-host one's Internet presence in suburban and rural America. The expectation is that one will use third-party services like Wordpress or Facebook, and now most of our infrastructure is built out to reinforce that trend. It's a little disheartening, to be honest.
I apologize for the inconvenience and the unexpected length of this downtime. I hope this will never happen again, and in case it ever does, I plan to build out some failover infrastructure in AWS. I have an account on Station^, where I try to post updates about my capsule status when it's offline.
Thank you for your understanding!
---
[Last updated: 2023-05-06]