nocrux

Nocrux is an easily configurable daemon manager that can be used by any user on the system. It uses double-forks to transfer cleanup responsibility of daemons to the init process.

• Synopsis
• Requirements
• Installation
• A note about child processes
• A note about managing daemons under a different user
• Changelog

Synopsis

usage: nocrux [-h] [-e] [-l] [-f] [--sudo] [--as AS_] [--stderr] [--version]
              [daemon] [command]

  Nocrux is a daemon process manager that is easy to configure and can
  operate on the user- or root-level. The nocrux configuration syntax is
  similar to Nginx. All users configuration file is in ~/.nocrux/conf,
  except for the root user, which is in /etc/nocrux/conf.
  
  This will start your $EDITOR to open the configuration file:
  
      $ nocrux -e
  
  The main namespace has the options and default values below:
  
      root ~/.nocrux/run;
      kill_timeout 10;
  
  You can also include other files like this (relative paths are considered
  relative to the configuration file):
  
      include ~/more-nocrux-config.txt;
  
  To configure a new daemon, you start a `daemon` section, specify the name
  and then the daemon's options in the news scope.
  
      daemon jupyter {
        cwd ~;
        run jupyter notebook;
      }
  
  You can now start the daemon with:
  
      $ nocrux jupyter start
      [nocrux]: (jupyter) starting "jupyter notebook"
      [nocrux]: (jupyter) started (pid: 10117)
  
  The following commands are available for all daemons:
  
    - start
    - stop
    - restart
    - status
    - pid
    - cat
    - tail
  
  You can specify additional commands like this:
  
      daemon jupyter {
        cwd ~;
        run jupyter notebook;
        command uptime echo $(($(date +%s) - $(date +%s -r $DAEMON_PIDFILE))) seconds;
      }
  
  Now to run this command:
  
      $ nocrux jupyter uptime;
      3424 seconds;
  
  Here's a daemon configuration with all available options and the
  respective default or example values:
  
      daemon test {
        # Example values:
        export PATH=/usr/sbin:$PATH;
        export DEBUG=1;
        run ~/Desktop/mytestdaemon.sh arg1 "arg 2";
        cwd ~;
        command uptime echo $(($(date +%s) - $(date +%s -r $DAEMON_PIDFILE))) seconds;
        requires daemon1 daemon2;
  
        # Options with their respective defaults:
        user me;
        group me;
        stdin /dev/null;
        stdout $root/$name.out;
        stderr $stdout;
        pidfile $root/$name.pid;
        signal term TERM;
        signal kill KILL;
      }

positional arguments:
  daemon        The name of the daemon.
  command       A command to execute on the specified daemon.

optional arguments:
  -h, --help    show this help message and exit
  -e, --edit    Edit the nocrux configuration file.
  -l, --list    List up all daemons and their status.
  -f, --follow  Pass -f to the tail command.
  --sudo        Re-invoke the same command with sudo.
  --as AS_      Run the command as the specified user. Overrides --sudo.
  --stderr      Choose stderr instead of stdout for the cat/tail command.
  --version     Print the nocrux version and exit.

Requirements

• Unix-like OS (tested on Ubuntu 15-17, Debian Jessie, macOS Sierra)
• Python 3.4+
• Node.py (optional)

Installation

$ pip3 install --user nocrux    # or
$ nodepy-pm install git+https://github.com/NiklasRosenstein/nocrux.git@v2.0.3 --global

A note about child processes

Nocrux can only send SIGTERM or SIGKILL to the main process that it originally started. If that process spawns any child precesses, it must take care of forwarding the signal! The thread Forward SIGTERM to child in Bash contains some information on doing that for Bash scripts. For very simple scripts that just set up an environment, I recommend the exec approach as described in the link.

A note about managing daemons under a different user

Example:

daemon gogs {
  user gogs;
  cwd /home/gogs/gogs;
  run ./gogs web;
}

If you're trying to manage a daemon that will be started by nocrux under a different user, you need the permissions to do so. For example, the superuser is allowed to do so and using nocrux as root should work immediately.

However, if you are not already the root user, nocrux will by default try to re-run itself as the user specified in the daemon, eg. in this case:

sudo gogs NOCRUX_CONFIG=/home/niklas/.nocrux/conf /home/niklas/.local/bin/nocrux gogs start

This will only work if

1. You installed nodepy system-wide, or you installed it with Node.py and the gogs user can read the path of the nocrux executable and the nocrux package directory
2. The gogs user can read your nocrux configuration file

Otherwise, you may be greeted with one of the following error messages:

• sudo: unable to execute /home/niklas/.local/bin/nocrux: Permission denied
• pkg_resources.DistributionNotFound: The 'nocrux==2.0.3' distribution was not found and is required by the application
• ModuleNotFoundError: No module named 'nodepy'

Changelog

v2.0.3

• Update for Node.py 2
• Add --sudo command-line option
• Add --as <user> command-line option
• Add support for variable substition in the daemon { export; } field
• Add support for custom signals for termination and killing a daemon process (see issue #21)
• Add support for custom daemon subcommands that have access to the following environment variables: $DAEMON_PID, $DAEMON_PIDFILE, $DAEMON_STDOUT, $DAEMON_STDERR (see issue #22)
• Add support for daemon{ root; } field which will change the parent directory of the default paths for the PID and standard output files
• Add support for # comments in the configuration file
• Change behaviour of daemon { user; } option, now serves as a default value for the --as option
• Fix configuration loading (daemon { run; } may now be preceeded by any other option)
• Fix -e, --edit now opens the editor always for the user's file

v2.0.2

• fix nocrux version command
• add nocrux edit command
• order of daemons when referencing them with all is now sorted alphabetically

v2.0.1

• removed fn:out, fn:err and fn:pid commands (actually already removed in 2.0.0)
• the default root config value will now be /var/run/nocrux if the configuration file is loaded from /etc/nocrux/conf
• more sophisticated config file parsing with nr.parse.strex module
• update error message hinting to check output of nocrux <daemon> tail if daemon could not be started

v2.0.0

• cli is now nocrux <daemon> <command> (switched)
• to specify multiple daemons, the <daemon> argument can be a list of comma separated daemon names
• configuration file is no longer a Python script
• configuration file must now be located at ~/.nocrux/conf or /etc/nocrux/conf
• nocrux can now be installed via Node.py
• add support for defining per-process environment variables

v1.1.3

• update README.md (corrected example and command-line interface)
• remove unusued -e, --stderr argument
• fix setup.py (use py_modules instead of the invalid modules parameter)
• enable running nocrux.py directly without prior installation
• add pid, tail, tail:out and tail:err subcommands

v1.1.2

• add setup.py installation script, remove nocrux script
• update README.md and renamed from README.markdown

v1.1.1

• close #18: Automatically expand prog ~ before starting process
• fix #17: PID file not deleted after daemon stopped
• close #16: Tail command is counter intuitive
• update output of command-line program
• process exit code is now printed to daemon standard error output file
• fixed stopping multiple daemons when one wasn't running
• implement #10: daemon dependencies

v1.1.0

• Renamed to nocrux
• Update README and command-line help description

v1.0.1

• Add krugs tail <daemon> [-e/-stderr] command
• Add special deaemon name all
• Fix krugs restart command