Issue with auto failover with unresponsive host

We have 2 DB instances: a primary instance host1:5432 and a read replica host2:5432

Here is my config:

client = postgres({
            host: "host1:5432,host2:5432",
            database: "postgres",
            username: "",
            password: "",
            target_session_attrs: "prefer-standby",
}

What I found is that it does not auto failover to the primary when the read replica can not be connected (Host exist but not respond). It will just print out error write CONNECT_TIMEOUT host2:5432 and stuck

same issue with primary

client = postgres({
            host: "unresponsive-host:5432,host1:5432",
            database: "postgres",
            username: "",
            password: "",
            target_session_attrs: "primary",
}

it will suck on error write CONNECT_TIMEOUT unresponsive-host:5432

Auto failover seems only work if host does not exist or host refuse connection

No source file, test, or entry point is named in the report. Reproduce the two configurations with an unreachable host and inspect the connection timeout and host-selection flow; done means the client proceeds to the eligible primary or standby instead of remaining stuck on CONNECT_TIMEOUT.