Ymir Report #22 — AWS outage

publishedover 1 year ago
2 min read

Heya friend!

Carl here. You signed up to receive updates about Ymir, the WordPress serverless DevOps platform that I’m building.


The big story this cycle is the major outage that AWS suffered last Tuesday. This was the first test of how the architecture would behave if an event like this happened.

The good news was that every customer's site stayed up during the outage. I was really happy and tweeted a bunch about it! I'll talk more about it in the marketing section.

Otherwise, this cycle felt more normal after the rough last two cycles. Still struggling to balance everything. But the migration is over for now, so it feels more manageable.


You can always view the history of Ymir's product development at https://ymirapp.com/changelog.

I'm back working on the Ymir product! Most of the work right now revolves around implementing features to the Ymir plugin that were problematic during the migration.

First, the plugin will now dynamically fix asset URLs. The issue was that, if you linked to assets in a post, those links would eventually break when you deployed new versions. The plugin will now check your post content and fix these URLs for you.

The second feature is having the plugin manage CloudFront cache invalidation. I have this feature almost wrapped up. The benefit of this is that I'll now be able to cache content a lot longer on CloudFront.

This will make Ymir-managed WordPress sites even more performant. It'll also increase their resilience if AWS has an outage like it happened this week.


The goal for this marketing cycle was to write a guide on how to do page caching with CloudFront. Things got a bit out of hand as they often do when I write guides lol. The whole thing got pretty massive. It's currently over 2000 words and I'm not finished!

I also lost a day during the cycle because of the AWS outage. I spent most of the day trying to figure out what was happening behind the scenes. But I also tweeted a lot because all the Ymir managed sites stayed up!

twitter profile avatar
Carl Alexander
Twitter Logo
December 7th 2021

I also did a Q&A video about the outage. In it, I talk a bit about the outage and why I think the Ymir sites stayed up while the main Ymir Laravel application went down.

This was before AWS released their report on the incident. It's pretty dry and technical, but the TL;DR is that they DDoSed themselves. But some services relied on EC2 APIs and they went down. So I wasn't too wrong in my guess in the video.

One thing that fell through the cracks and I want to revisit in the new year is to do a webinar. I'd done the poll. It did fine and I feel I should give it a shot.


You can always view Ymir's up-to-date business metrics at ymirapp.com/open. They're updated every 10 minutes.

Business is pretty active this month! I lost one customer, but I’ve had three trials this month so far. Two seem to have converted. I’m not sure about the last one yet.

Still talking to a few people about partnerships or ways to plant seeds with Ymir.

Otherwise, my main business issue at the moment is that my accountant ghosted me for over a month. I've dreaded dealing with accountants since this whole journey started. And even with them being referenced to me, and doing a discovery call and so on, they still ghosted me.


Hi! I'm a Creator.

Check out the resources I offer below and sign up for my newsletter!

Read more from Hi! I'm a Creator.