15b4da75-0c8b-4bb0-8f04-8d4167e64752

I understand that the app feels slower, but concrete numbers would be helpful for actually diagnosing the issue. As this is a new app, you don't necessarily have a direct comparison to your old app.

Please send logs with RTT from a user, as well as internal response times in the app, i.e. how long the request takes to reach the user vs how long the service takes to actually process the request

Breaking down even further would be even more helpful. RTT for communication between services & databases would be ideal

Actually i can compare both the apps directly, as the new one is a subpart of the old one. So basically, it should actually perform better than the old one

Anyways, I'll send you the logs and timings in a few hours

Or if we could get on a call, it would be much better

1. Is your external Mongo database located in Singapore too?

2. If every millisecond mattered, you would host Mongo on Railway and connect to it via the private network.

3. Your Redis database is in a separate project; it needs to be in the same project, and you need to connect to it via the private network.

^ Brody is able to see your project as he is a member of the team, take all this advice!

I'm using Mongo for another service also, so I'm just using a cluster on Atlas. And yes, redis is a good catch. I'll try that out and let you know

I think you should take point #2 into consideration, having Mongo in the same data center on Railway would significantly reduce latency, and you did say every millisecond matters.

No, actually I just checked, this service just authenticates once (which requires mongo). Other than that, everything else runs over websockets (not even requiring redis involvement). The redis task is running synchronously in the side, not even creating any trouble for websockets

Wait, I'll actually provide you with the other service's project id
b7229d4a-9cb5-4e53-a3cf-be4923d369f5

You can have a look at it. The only difference you'll find in both the services would be just this metal edge tag under the networking section

Can you manually migrate this from metal edge back to gcp? Maybe I can test it out and actually see if it makes a difference

So basically, what happened now is that i generated a new domain in my old service, which also migrated my old service to metal edge automatically. And now guess what, I'm getting the same lag as i'm getting on my new service

It definitely has something to do with metal edge

@Brody

How are you measuring latency? Could you also provide both of the measurements you're talking about?

I'll provide you the measurements in some time

what happened to their roles <:Thinking:1360710341239242762>

can they still chat here without the roles?

they can't because to talk here you need the support access role 😔

and they need the community access role to talk in #🎤｜chit-chat

my assumption is that the user left and rejoined and never went through sign-up again