I started being unable to connect this morning after no issues for weeks. Just checking that the server is up and it is an issue on my end - probably the firewall starting to block the traffic. If so I will need to only test at the house.
We are hosted on Railway, and I saw they have posted a service announcement: We're investigating issues impacting deployments - Incident details - Railway - Status
This began about an hour ago, according to that page. I just checked, and at the same time, our logs show an issue - likely caused by their outage.
I apologise: I am working on understanding their outage now.
1 Like
They appear to have had multiple services shut down: their support threads are full of dozens of people whose services have terminated. Ours did the same. I have manually restarted, and will monitor their status page.
When we go officially live we should have a status monitoring / heartbeat page, I think, and we have also discussed running on our own service hardware, not Railway.
(I’ve used them for a couple of years and run multiple sites with them, and really liked them; they are the new Vercel or Heroku. But recently failures have occurred multiple times. They just raised $100M in a Series A and I hope they will put that towards service reliability.)
Here is a HN thread on the outage: Railway (PaaS) global outage | Hacker News
And their post-mortem: https://blog.railway.com/p/incident-report-february-11-2026
They’ve been pretty open on the HN thread, which I appreciate – I personally really dislike the corporate approach where you just don’t know what’s going on, or there’s too much responsibility-avoidant language. It looks like they have an anti-fraud system, and they rolled out an update and it started taking down legitimate services.
There are lessons for us here too. All our other infrastructure is hosted on a different system, Azure, and CodeBot is here because I used Railway in the past and have been a fan of theirs, I find it an immensely useful and easy service, and so from the beginning I just shipped it there ‘because it worked’. Deployment and setup is incredibly easy and it freed me to work on coding the app, not worrying about servers. But our other infrastructure is more reliable, far more reliable (we track stats of course) and we should move CodeBot over.
When CodeBot is a paid service, outages like this would not be okay. IMO they are not okay even for a beta. I am sorry, and I also recognise that as the person who provisions and deploys CodeBot I am responsible for its uptime.