Improve coordinator port config

[phil-opp]

• Allow setting custom control port in coordinator
• Set default in CLI parsing -> default shown in --help output
• Add support for custom coordinator control port numbers in CLI

Follow-up to #513

[Gege-Wang]

Now, we can do this.

• To make coordinator_control_port configurable.

# remote
ubuntu@ip-172-31-30-172:~/dora$ dora coordinator --port 6014 --control-interface 0.0.0.0 --control-port 6013
Listening for incoming daemon connection on 6014

# local
$ ./target/release/dora check --coordinator-addr 54.234.111.197 --coordinator-port 6013
Dora Coordinator: ok
Dora Daemon: not running

• Daemon connect to remote coordinator

# remote
ubuntu@ip-172-31-30-172:~/dora$ dora coordinator
Listening for incoming daemon connection on 53290

# local
$ ./target/release/dora daemon --coordinator-addr 54.234.111.197:53290

$ ./target/release/dora check --coordinator-addr 54.234.111.197           1 ↵
Dora Coordinator: ok
Dora Daemon: ok

[Gege-Wang]

What does it do when we use dora up --coordinator-addr?
When coordinator and daemon are running in remote machine, dora up --coordinator-addr <remote-ip> in local machine doesn't seem to do anything.
When only a coordinator is running in remote machine, the local dora up --coordinator-addr does not seem to be able to start the connection between daemon and coordinator.

# remote 
ubuntu@ip-172-31-30-172:~$ dora coordinator
Listening for incoming daemon connection on 53290

# local
$ ./target/release/dora up --coordinator-addr 54.234.111.197              2 ↵
started dora daemon
daemon not connected after 2.1000001s

Did I misunderstand anything?

[XxChang]

What does it do when we use dora up --coordinator-addr? When coordinator and daemon are running in remote machine, dora up --coordinator-addr <remote-ip> in local machine doesn't seem to do anything. When only a coordinator is running in remote machine, the local dora up --coordinator-addr does not seem to be able to start the connection between daemon and coordinator.

# remote 
ubuntu@ip-172-31-30-172:~$ dora coordinator
Listening for incoming daemon connection on 53290

# local
$ ./target/release/dora up --coordinator-addr 54.234.111.197              2 ↵
started dora daemon
daemon not connected after 2.1000001s

Did I misunderstand anything?

It seems like that start_daemon() use default command which is dora daemon, So querying remote coordinator session will be foiled

[XxChang]

In addition, we should replace

cmd.spawn()

by

    if !cmd.status().wrap_err("failed to run `dora daemon`")?.success() {
        eyre::bail!("failed to start dora daemon");
    }

to capture cli error

[phil-opp]

Thanks for testing!

What does it do when we use dora up --coordinator-addr?

The coordinator-addr is currently only used for checking the connection after starting. So this doesn't really work (yet?).

For now, it's probably best to remove the coordinator-addr argument from dora up and instruct users to use the manual spawn commands for remote deployments. The commands dora coordinator for running the coordinator and dora daemon for running the daemon.

Edit: I just pushed 02b5c7d to remove the coordinator_addr and coordinator_port arguments from dora up.

[phil-opp]

In addition, we should replace

cmd.spawn()

by

    if !cmd.status().wrap_err("failed to run `dora daemon`")?.success() {
        eyre::bail!("failed to start dora daemon");
    }

to capture cli error

This would wait until the daemon exits, which we don't want to do here (we want to spawn it in the background).

[XxChang]

In addition, we should replace

cmd.spawn()

by

if !cmd.status().wrap_err("failed to run `dora daemon`")?.success() {

    eyre::bail!("failed to start dora daemon");

}

to capture cli error

This would wait until the daemon exits, which we don't want to do here (we want to spawn it in the background).

Copy that, i read it wrong, sorry

[haixuanTao]

Looks great thanks!