Running and managing two daemons with same executable by rc together

Question

I use FreeBSD UNIX and its rc system of startup scripts to make my Node.js servers start at reboot as a daemon and start, as a daemon again, and stop them with some commands the system gives me after I logged in to my FreeBSD UNIX user account.

I have read the Handbook related sections and an article on it, and I have written the following script, and the config, in /etc/rc.d/my_script and /etc.rc.conf, respectively. Which make it work as I described. When I add another script for another service only one of them work, I think usually the second one. Each server coded to listen to different port numbers.

my_script_1 is executable for all as '-rwxr-xr-x'.

/etc/rc.d/my_script_1:

#!/bin/sh

. /etc/rc.subr

name=my_service_1
rcvar=${name}_enable

project_dir="/usr/home/ec2-user/dev/projects/my_project_1"

command="/usr/local/bin/node"
command_args="${project_dir}/index.js"

load_rc_config $name
run_rc_command "$1"

/etc.rc.conf includes enabling var set:

my_service_1_enable="YES"

When I add another script and rc.conf entry like those with just name changed to a different name like, my_service_2, only one of them works. my_script_2 is executable as well.

What am I missing? How can I make them work both?

Answer 1

When you write "usually" it sounds like a race condition. It is then random which of the instances get to survive.

There is no basic problem in what you are trying to do with the rc scripts. The rc scripts only start the commands. The things you try to start does however need to be able to run in multiple instances. The rc system does nothing (as such) magically to do that.

From what you describe I would guess that parts of project 1 & 2 are using the same resources so node.js can only run one instance. An example would be that both projects bind to the same port. Two instances cannot serve on the same port - so only the first to start (the race) gets to live.

To debug this - then start by ensuring that you handle multiple instances properly. Try disabling your startup scripts. Then start both projects manually at the same time.

First you run:

/usr/local/bin/node /usr/home/ec2-user/dev/projects/my_project_1/index.js

And then:

/usr/local/bin/node /usr/home/ec2-user/dev/projects/my_project_2/index.js

As node outputs to STDOUT and errors goes to STDERR it is easier to see what goes wrong. I am pretty sure that your problem lies around here and that the output can help you further your investigation.

If the above actually works (contrary to my suspicion) then you need to get the log output from the starting node processes in your startup script. It does not seem like node.js logs to a file or syslog. This means that if you try to start the rc script manually you should get error output to the screen. So try controlling your rc script using:

service my_service_1 start

To look further into logging have a look at where is the nodejs-log-file.

NOTE: You are playing with your own scripts in /etc/rc.d/your_script which is intended for the base system. You should put your scripts in /usr/local/etc/rc.d/your_script. The service command searches both locations so it will work the same. No changes nedded to /etc/rc.conf either.