Scheduled for February 18, 2026 at 11:00 AM – 1:25 PM
Affects
Bot
Under maintenance from 11:00 AM to 1:25 PM
Website
Under maintenance from 11:00 AM to 1:25 PM
Backend
Under maintenance from 11:00 AM to 1:25 PM
Aether
Under maintenance from 11:00 AM to 1:25 PM
Completed
February 18, 2026 at 1:25 PMCompleted
February 18, 2026 at 1:25 PMMaintenance has completed successfully. There are a few issues with certain features/tooling but they're minor enough that I'm willing to let them be for now and will work on them within the next few days
Update
February 18, 2026 at 12:05 PMUpdate
February 18, 2026 at 12:05 PMEverything should be up and running now with all DNS records recreated with their new values. Some users may experience issues due to delays in DNS propagation and SSL certificate provisioning but these should eventually resolve themselves
Extensive review & monitoring will be taking place for the next hour or so to ensure nothing has been missed and to do some cleanup of older resources
Update
February 18, 2026 at 11:28 AMUpdate
February 18, 2026 at 11:28 AMFire is now back online and now running in Kubernetes!
Next up will be reconfiguring DNS records which may take some time due to SSL certificate provisioning. This means that web services will still be unavailable for a while
In progress
February 18, 2026 at 11:00 AMIn progress
February 18, 2026 at 11:00 AMMaintenance is now in progress
Planned
February 18, 2026 at 11:00 AMPlanned
February 18, 2026 at 11:00 AMDuring this time, we will be performing upgrades to Fire's infrastructure, changing how the bot and related applications are deployed/managed on the machine. These upgrades have already been tested in a production-like environment by moving Fire Beta to a separate machine for testing & learning.
The main upgrade should realistically only take 15-20 minutes with most of the time being spent replacing updating configurations for services exposed to the web as they can no longer be simply pointed to localhost & some port and even then, I don't expect this to take the entire duration of the maintenance but I'd rather schedule a larger window just in case
For anyone interested, I am moving Fire away from running directly on the machine and being managed via PM2 to using Docker containers & Kubernetes. This move will make it much easier to keep Fire up to date as upgrading the OS or Node version is as simple as changing the image used. It also comes with a more maintainable solution for offsite backups as I've configured the Postgres cluster with automatic backups to CloudFlare R2 rather than moving them to my Mac every so often
