Is it normal to frequently get "Super long ping" between two nodes on localhost?

[kanru]

Hi, for my project I'm planning to use the cluster feature. Right now I'm testing on localhost only.

I observe that quite frequently on my laptop that is plugged in power (so not in power-save mode), I sometimes see two totally idle nodes spit out

2025-01-09T00:40:27.371595Z  WARN Actor{id="0.0" name="supervisor"}:Actor{id="0.1" 
name="node"}:Actor{id="0.2"}:Actor{id="0.3"}:Actor{id="0.4"}:Actor{id="0.6"}: ractor_cluster::node::node_session: 
Super long ping detected 127.0.0.1:8001 - 127.0.0.1:42746 (77ms)

And if there are moderate activity between the nodes (rpc call every ~250 ms), the super long ping message appears more frequently.

HW: AMD Ryzen 5 7640U (6 cores 12 threads)
Runtime: tokio rt-multi-thread

[kanru]

Actually with the debug log on, I can see the ping latency is consistently around 40 ms between two idle nodes. I'll investigate more as it seems something fishy is going on.

[kanru]

The 40 ms delay probably comes from the infamous TCP Delayed Acknowledgement. The default delack timeout on Linux is 40 ms.

[kanru]

Opend PR #319