a month ago
A bunch of services are showing problems.
Some servers go to sleep and don’t wake up after access.
Some deployments never sleep, even with the sleep option enabled.
I need someone to fix this. It’s consuming my money because the services keep running nonstop, and when they do sleep, they don’t work anymore unless I redeploy.
Honestly, I’m already checking other options. Railway isn’t competitive anymore.
Attachments
2 Replies
a month ago
According to Railway docs:
For Railway to put a service to sleep, a service must not send outbound traffic for at least 10 minutes. Outbound traffic can include telemetry, database connections, NTP, etc. Inbound traffic is excluded from considering when to sleep a service.
Your service might be sending some outbound traffic, which prevents them from going to sleep.
Slept services still consume a slot on Railway's infrastructure, enabling Serverless de-prioritizes your workload and in remote cases, may require a rebuild to re-live the service.
That should be pretty self-explanatory. Some instances may require a rebuild to revive the service.
a month ago
Just to add on, outbound does not just mean egress. I had a rails app pinging a postgres database every 1 min for background jobs. Its a small app, so i moved the background jobs to a local sqlite database. This stopped the constant database chatter and now my services will sleep.
What makes it tricky is that the http logs that show traffic is really just egress traffic, not internal network traffic. So you have to debug it yourself. Maybe the Network Flow Logs show internal traffic, i'm still learning those.