reactor_netty_connection_provider_active_connections is always 0, despite successful http requests

Hello Reactor Team,

I wanted to repot an issue where the metric reactor_netty_connection_provider_active_connections always stays at 0, despite successful http requests.

Setup of experiment, at the hour, there is a trigger from an external app to my Spring Webflux app (I am the server). My Spring Webflux app will then make an http request (outbound, this time, I am the client) to a third party API.

The third party API is very resilient, and never fails. For proof, we do always get their HTTP response back.

Now, looking at the four metrics described here: https://projectreactor.io/docs/netty/release/reference/index.html#_metrics_3

reactor.netty.connection.provider.total.connections
reactor.netty.connection.provider.active.connections
reactor.netty.connection.provider.idle.connections
reactor.netty.connection.provider.pending.connections

I am surprise why active stays at 0, always.

I do see total.connections being incremented at every hour. I would say this is expected, since every hour, I make a successful http request to someone else.

I do see idle.connections being incremented at every hour. I would say this is expected, since every hour, I make a successful http request to someone else, I would imagine, after the successful request, the connection becomes idle, and stay that way forever

I do see pending.connections being flat at 0. but I would say this is expected since the http request is always successful, nothing is waiting, nothing is pending.

But I am surprised active stays at 0. I would have expected, at some point, active being incremented, since there is some kind of connection between me and the third party app.

Is it possible there is an issue this particular counter on not being incremented please?

Thank you

[patpatpat123]

Hello @violetagg,

Indeed, this is working, and I am now seeing the metric being incremented. Thank you

[patpatpat123]

Understood, and thank you for pointing me to this, I will try to investigate if it is not a “me issue”.

Since there is a set of test already in place, maybe it is just me.

On my side, I am just also raising the issue based on Grafana data:

For example here, we can clearly see http connections, with address being resolved, connect and SSL handshake.

But is those graph, we can clearly see reactor.netty.connection.provider.active.connections (the graph on the left) was flat 0 all the time. I was expecting an increase matching the time stamp.

Will dig further