when using <https://ktor.io/clients/http-client/features/timeout.html> and `requestTimeoutMillis` it...

# ktor

[rrva]

when using https://ktor.io/clients/http-client/features/timeout.html and requestTimeoutMillis it is not needed to set connectionRequestTimeout on the engine level as well, right?

I am thinking to add to the following code

val client = HttpClient(Apache) {
    engine {
        connectionRequestTimeout = 400
    }
    install(HttpTimeout) {
        connectTimeoutMillis = 400
        socketTimeoutMillis = 100
    }
}

into

val client = HttpClient(Apache) {
    engine {
        connectionRequestTimeout = 400
    }
    install(HttpTimeout) {
        requestTimeoutMillis = 400
        connectTimeoutMillis = 400
        socketTimeoutMillis = 100
    }
}

Mostly to cover time spent in cases not covered by the other three timeouts (name resolution in DNS for example)

[e5l]

Hi @rrva, yes. The engine level configuration is obsolete since we have a common timeout feature

[rrva]

so it’s recommended to only set requestTimeoutMillis and nothing more?

[Cyrille QUÉMIN]

This is not true for the CIO engine. Or not true anymore perhaps? as I have posted today a question around this. Since I can see that both timeout are used but I can't really understand exactly when was is vs the other one