N/A

My Usage Seems Fine , I'm on the Pro Plan

tell me more about the server you run on railway

It takes in a bunch of images and audio file and uses ffmpeg to convert the images and audios into one video file

what's the tech stack

the server is NodeJs and the Client is NextJs

how many replicas do you currently have

Just one

It works fine on my end and other users have no issues too most of the times. But occationally the error pops up

try adding a few more replicas

How many more would you suggest? And why does that solve the issue if you dont mind

start with 3 total,

a single node instance may not be able to handle all of the traffic.

But shouldnt i see the resourses maxing out if thats the case?

a single node instance cannot use more than 1 vCPU since node is single threaded

Fair. After adding the replicas do i need to change any code or the allocation will be automatic?

we will run 3 instances of your app and handle the routing, as long as you don't store user state in memory you should be fine

You recommed i add 3 different locations or use a single one

that's up to you, do you have users from different locations?

wouldn't want to add replicas in locations you don't have any users

I do have users from different locations. Mostly Us/Europe , but i'll experiment with both and compare

If the issues persist i'll let you know

Thank you!

sounds good!

So i don't know if this issue is related

But after turning on multiple replicas now , my server side events are not working

Usually i have SSE set up so that the progress of the video generation is sent from the nodejs server to the client constantly. But now the SSE does not work at all after turning on mutiple replicas

The server works fine and the video is generated successfully but the SSE updates does not show on the frontend. There is a browser error though

SSE does work with multi-region replicas, so this would be something with your code, what's the actual error? and are there errors in your deploy logs?

I'm using single region replicas. This is the deploy logs

you have replicas in multiple regions

those are old deploy logs

These are my replicas

My bad , there is no new deplog error logs

ah I was just going off of this, either way, SSE fundamentally works with a single region or multi region replicas

In that case the issue might be something else. I'll debug. It might just be a coincedence that it started right after i added replicas

i would say that you are storing state in memory

the replica that is doing the ffmpeg stuff may not be the the replica that answered the SSE request

I'm not sure if i am storing state in memory or not. I don't know what that exactly means , but i'm using SSE to track the progress and send to the client

yeah then you are doing ffmpeg stuff on one instance and another instance doesn't know about it so SSE fails, you need to store the progress of a ffmpeg job globally in redis so that all replicas have access to the state of any ffmpeg job

Hmmm thank you for that!

it's just a guess from what you've told me, I could be wrong, please still do your own research

I will. Thanks Brody

no problem!

!s