N/A

I would expect it to wait for the container to start up again and then display the page instead of just responding with a bad gateway

while the container is still being started

I assume that means the container did start but your app didn't respond in time

do you run migrations in your start command?

I do not

what kind of app is this?

nextjs site

which is not connected to any database or anything

just a static website

what's your start command?

Before the app starts railway responds

did not change it

not using custom

so its most likely just npm run start

are you using a health check

i am not

go ahead and implement that

just putting a / will work?

if it's a static site, yeah just the root path is fine

I can't remember if railway checks the health check when resuming a slept app, but it's good to have either way

Yeah it doesnt work

Before Starting Container appears in logs railway responds with a 502

i mean at the same time^

okay, i asked the team about this and will get back to you when i have more information

Thanks

did you need an immediate solution? because i may have something you can try

No

I will just disable app sleeping in the meantime Lol

sounds good!

got an answer back, this basically means your app was not able to answer the initial http request fast enough.

when there is an incoming request for a slept service, the sleeping service is started and is tcp probed every 30ms for a maximum of 10 seconds, once a prob is successful the http request gets sent to your app, if you app doesn't respond then you will see a railway error page

you have a few options -

• figure out why your app accepts a tcp connection but not an http request and remedy that.

• stick caddy in front of the service to retry connections on non-200 status code responses.

• keep app sleeping disabled.

We've resolved an issue where apps with longer startup times were showing 502 errors. Apps now have up to 10 seconds to start accepting traffic, thus preventing these error pages from appearing.

You will need to trigger a deployment so that the changes we have made take effect.

cc @Bartek