Frequent "Connection timed out" errors when calling Management API

[jonathan.rau]

Hi there!

We are building a service to manage users and organizations using the Management API. However, our calls to the Management API fail sporadically as the connection times out:

2021-09-27 10:23:57.562  WARN 1 --- [or-http-epoll-1] r.netty.http.client.HttpClientConnect    : [id:53e822b5-2, L:/172.17.0.2:60426 - R:<domain>.eu.auth0.com/104.18.244.88:443] The connection observed an error

io.netty.channel.unix.Errors$NativeIoException: readAddress(..) failed: Connection timed out

2021-09-27 10:23:57.564 ERROR 1 --- [nio-8080-exec-1] o.a.c.c.C.[.[.[/].[dispatcherServlet]    : Servlet.service() for servlet [dispatcherServlet] in context with path [] threw exception [Request processing failed; nested exception is org.springframework.web.reactive.function.client.WebClientRequestException: readAddress(..) failed: Connection timed out; nested exception is io.netty.channel.unix.Errors$NativeIoException: readAddress(..) failed: Connection timed out] with root cause

io.netty.channel.unix.Errors$NativeIoException: readAddress(..) failed: Connection timed out

(I have replaced the domain with a placeholder)

This usually happens, when the service has not been used for a while and subsequent requests are usually served in a reasonable time (~500ms). We are currently on the free plan and are trying out the organization feature. Are those timeouts related to auth0 using a serverless architecture in the backend, that needs to warm up? Can we expect the performance to be better when switching to a paid plan?

[konrad.sopala]

Hey there!

What you’re asking about our architecture is definitely not the reason. I would actually start researching this from diving deeper into why you have so high response times ~500 ms.

[jonathan.rau]

Thanks for the quick response. There is a small typo in the original post as the response times of the subsequent requests are below 500ms (< not ~). I apologize for that.

We’ve investigated our code a little further. By examining the log statements from today, we found out, that the client_credentials exchange is causing the timeout and not the call to the Management API. This might just be related to the current incident. So we will keep observing the response times in the upcoming days and I will come back to this issue when I have more insights.

[konrad.sopala]

Perfect! Let us know if you have any other questions down the road!