A transport can log a failed proxy connection while the application also receives and handles the reactive error. The log entry alone does not establish that the application's handler was skipped.

Follow the error signal in the chain that actually gets subscribed. A try/catch around code merely constructing a Mono is a different scope from handling a later network failure.

OperatorEffect
doOnErrorPerforms a side effect while the error continues
onErrorMapTransforms an error into another error
onErrorResumeSwitches to an alternate publisher

Choose the behavior the application needs. Recording a metric does not require returning fake successful content. An alternate response should be a deliberate application result, not a way to hide a failed route.

Also distinguish transport exceptions from received HTTP error statuses. A proxy connection can fail before an ordinary destination response exists, so an HTTP-status handler may have nothing to inspect. Preserve the cause of a WebClientRequestException when diagnosing the lower-level failure.

Adjust logging only after understanding the outcome. Turning the entire transport logger off can remove evidence needed for a later timeout, TLS problem, or authentication failure without changing the behavior of the reactive pipeline.

Ũhoro ũhĩtĩte