In my case, I believe you are right, though the thread is sitting here rather than specifically on multiselect.cfc:298 https://github.com/Ortus-Solutions/commandbox/blob/development/src/cfml/system/Shell.cfc#L398

yeah... it did get there by way of https://github.com/Ortus-Solutions/commandbox/blob/development/src/cfml/system/services/ServerService.cfc#L341 so, as you suspected, it is waiting on an answer about what to do because it thinks the server is already running when it isn't

Since 'server stop' doesn't throw an error when you try to stop a server that isn't running (and because the nssm startup directory is the web root), updating the nssm command from

server start --console --debug

to

server stop && server start --console --debug

seems to be working so far. After that update, I've done the following multiple times and the server has been coming back up consistently the first time • manually stopped and started the service using the Windows service manager UI • manually restarted the service using the service manager UI • used stop-service and start-service PowerShell commands • used restart-service PowerShell commands • rebooted the server without stopping the service first (because the service is set to auto start delayed) • rebooted after stopping the service first I'm going to repeat those tests a few more times each before calling it a successful workaround (provided you can't think of any negative impact I'm not seeing yet, @bdw429s).

The "real" way to avoid the prompt is to force the CLI into non-interactive mode like I said. I haven't found any reliable way to detect that within CommandBox itself, which is why I'm just setting the env var.

Let me know if you don't know how to set env vars with NSSM.

The service manager module would already do this for you 🙂

Or you're on an older version of CommandBox

preServerStart

fires right before the "is running?" check and I think that's when dotenv loads, so in theory it should be there

You didn't say where you ran that command

Relegating it to the service means it only applies then

Sorry, are you saying it should go in the start command that NSSM is using?

That's where it makes the most sense, since you really only want that flag set when specifically starting the server as a service

I should charge you to look at that though since it's from the Service Manager code 😉

You can basically give NSSM one or more env vars and it will toss them into the environment where it starts the service

Handy for this sort of thing

guessing not as long as its only being used on one server at a time

Which actually has some new coolness added to it

You can now also use it to create services out of Task Runners

task service create

I just build the task stuff this week (after wanting to do it for years) for a client project where we needed to spin up some Rabbit MQ consumer threads on a random windows server to run active directory commands.

I'm using the new RPC Server/client stuff I just built into the RabbitSDK so their docker/linux hosted app can still call out to our little micrservice running on WIndows that is tied to those native exe binaries for messing with their LDAP accounts

It just requires an RPC server running in my task runner which has a

while( true ) sleep( 1000 )

after it

channel = rabbitClient
	.startRPCServer( 
		queue='RPC_requests',
		consumer=(method,args)=>{
			return "hello world";
		}
	);

And then our app sends it a requests via message queues

result = rabbitClient.RPCClient( 'RPC_requests' ).sayHello();

It's super lightweight and prevented us from needing to build out an entire HTTP/MVC/REST API just to host like 3 little calls

RMQ events*

I have a number of task runner based daemons for this client (most of them running minibox docker on their swarm) that push real-time events to their dashboards all via Rabbit

is it this? https://www.rabbitmq.com/event-exchange.html

Ah, I see- you meant your experience was limited. Got it-- I thought you were saying it had limited features. 👍

Or if you just meant the normal thing sending a message to a queue, etc, etc

And when a websocket publishes a message, it just goes into a queue as well

nice

SO each browser authenticates to stomp with thei JWT and then I know • who they are • what channels they can subscribe to

Then we have custom topics for each user so i can push a message to

user_notifications.bobby

and it pops up in only your browser (if you're online)

I even built a REST over STOMP plugin that allows you to remote-execute coldbox events over stomp as an alternative to Ajax as it re-uses existing established TCP connection instead of re-negatiating.

Yeah, my client built their own basic "Slack" chat into their app

And tied it into Twillio so their agents can live message where the recipient is actually via text message

We've also converted a lot of their "every 5 minutes" sort of scheduled tasks that, say, look for a new incoming lead in the DB to just fire an async message right when the lead comes in and then a rabbit consumer running out in a Docker swarm task runner immediately processes it

We also rely heavily on Couchbase for super-fast caching of data that's been handled by our real time processes to reduce DB traffic

Lots of good stuff to pile onto my tinker list. Thanks! I'm going to go do actual work now I guess.

I've not been able to get the nonInteractiveShell option to stop the problem. I've set it through NSSM like you mentioned (and confirmed that its there when I edit the service). I also went into the webroot of the server and ran

config set nonInteractiveShell=true

(and confirmed it stuck with

config show

) no matter where I put it, it didn't stop the problem of the start hanging while waiting on input.

server status && server start ....

also didn't work as a replacement for

server stop && server start ...

it just got the whole thing caught in a loop as it filled up %commandbox_home%\.logs\server.out.log with these

[ERROR] runwar.context: java.net.ConnectException: Connection refused: connect
[ERROR] runwar.context: at java.base/java.net.PlainSocketImpl.connect0(Native Method)
[ERROR] runwar.context: at java.base/java.net.PlainSocketImpl.socketConnect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.AbstractPlainSocketImpl.doConnect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.AbstractPlainSocketImpl.connectToAddress(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.AbstractPlainSocketImpl.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.SocksSocketImpl.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.<init>(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.<init>(Unknown Source)
[ERROR] runwar.context: at runwar.Stop.stopServer(Stop.java:26)
[ERROR] runwar.context: at runwar.Server.startServer(Server.java:206)
[ERROR] runwar.context: at runwar.Start.main(Start.java:50)
[INFO ] runwar.context: Could not stop server.  Are you sure it is running, and listing for stop requests on port 49909?
[ERROR] runwar.context: java.net.ConnectException: Connection refused: connect
[ERROR] runwar.context: at java.base/java.net.PlainSocketImpl.connect0(Native Method)
[ERROR] runwar.context: at java.base/java.net.PlainSocketImpl.socketConnect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.AbstractPlainSocketImpl.doConnect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.AbstractPlainSocketImpl.connectToAddress(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.AbstractPlainSocketImpl.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.SocksSocketImpl.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.connect(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.<init>(Unknown Source)
[ERROR] runwar.context: at java.base/java.net.Socket.<init>(Unknown Source)
[ERROR] runwar.context: at runwar.Stop.stopServer(Stop.java:26)
[ERROR] runwar.context: at runwar.Server.startServer(Server.java:206)
[ERROR] runwar.context: at runwar.Start.main(Start.java:50)
[INFO ] runwar.context: Could not stop server.  Are you sure it is running, and listing for stop requests on port 50025?

the

server stop && server start...

workaround is holding strong though

it didn't stop the problem of the start hanging while waiting on input.

And you • confirmed the thread was specifically running the mulitSelect code? • Are on CommandBox 5.5.2? It should really be impossible to hit that code path when non interactive shell is turned on.

Here's the check for that

/**
	 * Is the current terminal interactive?
	 * @returns boolean
  	 **/
	function isTerminalInteractive() {
		// Check for config setting called "nonInteractiveShell"
		if( configService.settingExists( 'nonInteractiveShell' ) && isBoolean( configService.getSetting( "nonInteractiveShell" ) ) ) {
			return !configService.getSetting( "nonInteractiveShell" );
		// Next check for an Environment Variable called "CI" that is set
		} else if( systemSettings.getSystemSetting( 'CI', '__NOT_SET__' ) != '__NOT_SET__' ) {
			return false;
		// System.console() will return null is the stndin or stndout is not a TTY
		} else if( isNull( createObject( 'java', 'java.lang.System' ).console() ) ) {
			return false;
		// Default to true
		} else {
			return true;
		}
	}

That config setting trumps everything else

But again, it was added in the latest 5.5.2 update of CommandBox-- earlier versions of server start did not check if the shell was interactive or not before prompting the user

dang... no auto-upgrade between 5.5.1 and 5.5.2

I couldn't stand it... I upgraded and retested with JUST the env var (set through NSSM environment vars). Not only does that seem to have fixed the issue (as well), the server start up is actually much faster now.

I'll run a bunch more service stop/start/restart and reboot tests but I'm pretty confident the issue is gone.

...yes, I removed 'server stop &&' from the service start command 😉