Proxy metric outbound_tcp_errors_total

[0xttfx]

I found the transport layer metrics documentation (https://linkerd.io/2-edge/reference/proxy-metrics/#transport-level-metrics) but the information is limited for the métric "outbound_tcp_errors_total" metric

linkerd-stable diagnostics proxy-metrics -n prod po/api-catarse-6f776577c7-5cflx |grep -E "^[io].*bound_tcp_errors_total" ... ... outbound_tcp_errors_total{target_addr="34.203.110.86:5432",target_ip="34.203.110.86",target_port="5432",error="unexpected"} 1 outbound_tcp_errors_total{target_addr="34.203.xxx.xxx:5432",target_ip="34.203.xxx.xxx",target_port="5432",error="i/o"} 30070

In my case, I see that I have "i/o" errors! But what reading do I make of this result?
In my research, I found "error: i/o timeout" which allows me to understand the cause of the problem!

But just "error: i/0" is too generic...

Is there any other documentation about transport layer or "outbound_tcp_errors_total" metric?