We're evaluating Retool with gRPC APIs but we found it usually took more than 20s to get result from gRPC API. and accoding to server log, our backend receives reflection requests from Retool immediately, but waits 20 seconds and receives the gRPC APIs. Is this a known issue or are we missing something in the deployment.
Retool version: 3.10.2
We might find the root cause. One of our gRPC service have more than 300 APIs.
When we execute one of these API, it seems that Retool fetches the server reflection info of all 300+ gRPC APIs and CPU usage goes to 100% when Retool parsing these server reflection. And it always takes more than 20s+ to do the parsing job and then it sends the request to server.
I think it's a bug or an optimization that when user trigger a request, Retool backend doesn't need to parse all its API reflections.
Hey @yuntao, can you share how you have this set up? Are you using the
Connect to gRPC using Server Reflection option?
Yes, we're using server reflection API
Got it, thanks for that info @yuntao. I've filed a feature request to limit server reflection on query run. We'll update you here when there is any additional information to share.