Mind sending full build logs here anyway?

Use this tool to send it: https://bookmarklets.up.railway.app/log-downloader/

I remove docker file and try using only fastapi. railway.json {

"$schema": "https://railway.app/railway.schema.json",

"build": {

"builder": "NIXPACKS"

},

"deploy": {

"startCommand": "uvicorn app.main:app --host 0.0.0.0 --port $PORT"

}

}

Are those all the logs available?

Yeah

It just... stops?

Hey all, we're looking into this right now. Will update soon

As an update - the reason your build is failing is because the build container is OOMing when building your image. We're able to reproduce and have found that the new runtime significantly reduces memory usage and addresses the OOM. However, we're still investigating some transient errors so don't have a definitive answer quite yet.

That said, could we try enabling the new runtime on this service @ds-digital-23? We can switch it on from our side and troubleshoot your service directly, if you are okay with that.

I agree. However, I removed 2 large files (103MB each). These are two YOLO models. Could I add them back?

Before you add them back, want to try building with the V2 runtime enabled and see how that goes? if it continues to fail, to get you unblocked we can disable V2 builder. All of this is done in the service settings.

Since we're able to reproduce, we can continue working on a fix and let you know when to switch back.

Ok

Hey, I created another project and worked it! However https://enecad-api-production.up.railway.app/docs gets Application failed to respond.

For that, please see this docs page -

https://docs.railway.app/guides/fixing-common-errors

Amazing. Also FYI to both, we patched the issue that was causing the failures. Definitely please let me know if you see it again.

Hi! I get it. But now I got an error about ASGI. It’s not clear for me. Could you guys help me?

Please provide the full build or deployment logs -

https://bookmarklets.up.railway.app/log-downloader/

here

Is this a fastapi app?

That's right!

What's your start command?

CMD ["sh", "-c", "uvicorn main:app --host 0.0.0.0 --port ${PORT:-8080}"] -> Inside the dockerfile

Do you have a start command set elsewhere like in your service settings?

No

I tried to set only one YOLO model, but got error.

Hi! Somebody? The same code runs well in other platform. Anyone helps?

Where are you storing these model files? and how big are they?

Each one has 103MB (there are two YOLO models). They stored at "ia" folder.

Meaning they are pushed into GitHub's LFS.

Railway will not pull files from LFS, you will need to find an alternative place to store those model files.

Ah! Ok!

Thanks!