12 days ago
Hey community!
Could someone help with these, I am not sure if this is safe to ignore or symptomatic of something I should be worried about. Visitors on the site are fine, the build is clean and everything is working. I still received a deploy crash email from Railway, not sure what to do with it and what the root cause is.
Plan: Hobby. Region: EU West.
App: Next.js 16.2.6
Settings: Enable Serverless ON, no healthcheck / cron, Restart Policy On Failure (10).
I received a "deploy crashed" email on 2026-09-24 at 09:26 UTC that read something like this:
Deploy Crashed!
Uh oh. Your deployment for XXX in YYY crashed within the production environment.
It might help to look at the deploy logs and if you're still stuck and think we can help, feel free to reach out to us on our Discord.
Regards,
The Railway Team
I never received one before, although the service sleeps and wakes often
(64 "Stopping Container" events over 3 days on the previous deployment).
So I looked at the deploy log and here's some evidence (UTC timestamps, from deploy logs and HTTP logs):
- Last response before the stop: about 09:18:21. The stop came about 8 min later, so it looks like a normal idle sleep.
- 09:26:16.846 Stopping Container
- 09:26:17.320 the new process printed "Ready in 147ms", then "npm error command failed / signal SIGTERM" from the old one
- 09:26:17.865 Starting Container
- A request arrived at 09:26:17 and got 200 after 1440 ms, with one upstream "connection refused".
- So the container was stopped for sleep and restarted about 1 s later by an incoming request. The email matches this stop to the minute.
- A normal sleep does not email: a stop at 07:30:00 was followed by the next start at 07:40:34, with no email.
- The app never threw an exception, and memory stays under 400 MB.
- One stop does not fit sleep: 2026-09-24 01:16:17 came 9 s after a request (next start 01:16:48).
Questions:
- Is a sleep interrupted by a request within 1-2 s classified as a crash and emailed?
- What exactly triggers the "deploy crashed" email?
- Is there a setting to avoid crash emails for serverless sleep while keeping wake-on-request?
1 Replies
12 days ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • 12 days ago
12 days ago
I’d try this: https://docs.railway.com/deployments/troubleshooting/nodejs-sigterm-handling.
For 2: when your deployment literally crashes on a nonzero exit code.
For 3: You can disable email notifications for crashed deployments by going to https://railway.com/account/notifications and disabling the Deployment Crashed event. However, this will remove the notification all crashes, not just serverless services.