Connection closed error=connect timed out after 1s client

[Sul-ss]

We are facing Performance degradation after Injecting LinkerD and trying to figure out the root cause and what to enhance.

I notice this in linkerd-proxy
INFO ThreadId(01) outbound: linkerd_app_core::serve: Connection closed error=connect timed out after 1s client. INFO ThreadId(01) outbound: linkerd_app_core::serve: Connection closed error=connect timed out after 1s client.

Running this command I was able to get the target destination and ports.
linkerd diagnostics proxy-metrics -n namespace pod/pod-name | grep ^outbound_tcp_errors

Result:

outbound_tcp_errors_total{target_addr="xx.x.x.x:443",target_ip="xx.x.x.x",target_port="443",error="i/o"} 4
outbound_tcp_errors_total{target_addr="xx.x.x.x:55225",target_ip="xx.x.x.x",target_port="55225",error="i/o"} 24
outbound_tcp_errors_total{target_addr="xx.x.x.x:9999",target_ip="xx.x.x.x",target_port="9999",error="unexpected"} 100
outbound_tcp_errors_total{target_addr="xx.x.x.x:5672",target_ip="xx.x.x.x",target_port="5672",error="i/o"} 1
outbound_tcp_errors_total{target_addr="xx.x.x.x:443",target_ip="xx.x.x.x",target_port="443",error="i/o"} 97
outbound_tcp_errors_total{target_addr="xx.x.x.x:443",target_ip="xx.x.x.x",target_port="443",error="unexpected"} 50

the target destinations are HTTPS endpoints "internal to cluster" as well as external services like databases and RabbitMQ .

Can these timeouts cause performance degradation even though I don't see any errors in applications pods?

Adding config.linkerd.io/skip-outbound-ports: 55225,9999,5672 will remove the errors for the mentioned ports but not the HTTPS endpoints.

Note that Proxy resources seems more than enough.