Hey there!
For some info on this, we had an incident that caused a percentage of our deploys to go down.

That being said we're also looking into how to better notify users of impact when it happens. I don't have the best solution for you but in the future I'd hope we notify as soon as we detect it happening.

Can subscribe to our status page and get email alerts when we call an incident.

If you're interested the full post mortem on that is here
https://blog.railway.com/p/incident-report-february-11-2026

Thanks for the reply. Great to see elaborate postmortem. Any idea why automatic recovery did not work in my case?

I'm not quite sure entirely why it didnt work in your case, we should've restarted everybody that we got a log of. Theres a chance that a very small few were dropped in that so i apologize.

Out of maybe 10 services that were shut down, less than 50% of them were automatically restarted. And I have read in other threads as well that the restarts didn't work. So I am not sure how true that actually is.

Just because we attempt a restart doesn't mean that it will be fullproof. Example, an application with no reconnect logic will not reconnected to a SIGTERMed DB. Hence why you see the whole support team trying to scour for more affected cases.