OnlineOrNot updates from January 2024

Over December and January I focused on OnlineOrNot's public API, making it possible to see what OnlineOrNot saw when it detected a failing uptime check, as well as fixing bugs and cleaning up tech debt across all of OnlineOrNot (it's getting faster and faster to release updates).

Table of contents

• What's new
  • Adopting OpenAPI and publishing new API docs
  • OnlineOrNot, what did you see?
• Thank you

What's new

Adopting OpenAPI and publishing new API docs

OnlineOrNot's public API was originally written to serve a single purpose: provide data for the CLI. The documentation for the API was handwritten, and lovingly placed alongside OnlineOrNot's other hand-written docs.

This made it time-consuming to update the API: first I would write code, then remember to context-switch into "content-mode" and decide where the new API endpoint docs should live, then actually write docs from scratch, and finally deploy the content.

In December, I automated this process (I wrote more about the technical details here), and now we now have a published OpenAPI schema and a new API documentation hub:

The API itself (like OnlineOrNot) is still under active development, and new endpoints will be automatically published to both the documentation site, and the OpenAPI schema in the coming weeks.

OnlineOrNot, what did you see?

Up until now, OnlineOrNot would only ever tell you if your website or API was up or down, and roughly how long your website or API took to serve a request. Actually figuring out what was wrong was "left as an exercise for the reader", like an academic textbook.

As a user of OnlineOrNot, you would have one of two reactions to this:

1. I better investigate my logs around that time, see if the database backup job isn't accidentally killing our website at the same time.

2. Nah, that's a false alarm, no way the website goes down at exactly 2am every single day.

To remedy this, in January I focused on making every request result (HTTP status code, region it checked from, response time, date and time checked) available in the dashboard, as well as the request/response metadata (request headers, response headers and response body).

Every uptime check in OnlineOrNot now displays a list of recent checks:

with an option to view all checks, and filter by failing checks:

and finally, the raw metadata involved in the check:

By the time you see these screens in OnlineOrNot, they will likely be even more featured - it was just so useful that I wanted to release and share an early preview as soon as possible.

In short, performing a root-cause analysis of why an uptime check failed with OnlineOrNot is now significantly easier.

Thank you

OnlineOrNot only exists because of folks like you.

I don't have a massive marketing budget, OnlineOrNot gets its users from folks like you enjoying it enough to tell your friends and colleagues about it.

So here's an ask: I work on the things that people care enough about to tell me about them. If you think there's something missing, or something that needs improving, reply to this email/email me at max@onlineornot.com - I'd like to know more! You can even book time in my calendar.