Hello guys, how's it going? I'm having an issue with the read relationships api. I've been getting some 504 Gateway Timeout every now and then. Is there a case in which spicedb would throw a 504? I'm using the http api by the way. Can't find any logs on spicedb either

Are you running NGINX or Envoy as a reverse proxy in front of SpiceDB?

I believe so. That's probably it, isn't it?

either that or you're not using pagination

Hm, that could be it too. I'm not paginating

I'd recommend using pagination anyway

Is there pagination on the read relationships api? Can't find an example

the http one mirrors the rpc

@Joey , if the number of relations is super large, would I also get a 504 gateway timeout? Since it's a stream, wouldn't it be another error?

not if called via the rest gateway

I'm wondering, could these errors be from the grpc gateway? If so, how could I debug it?

its unlikely unless the request is timing out

I understand. But I still don't think that's what's causing this issue because some of these timed out requests work fine when calling from my machine, and they often don't even return any relationship. Spicedb metrics such as response time are also fine. We couldn't find error logs on spicedb or on our istio side cars either. I'll keep digging. Is the grpc gateway another application or does it run inside spicedb?

is this against Authzed serverless or your own install or...?

We are running it on our kubernetes cluster

the REST gateway in that case is run in the same process, but its "fronts" the GRPC API

So it's unlikely the problem indeed