The babysitter is meant to alleviate the tedium of starting & managing multiple processes when developing distributed applications. It has been used daily to manage 30+ simultaneous processes with negligible overhead. This manager is meant to be used FOR DEVELOPMENT ONLY. Please use something better for production such systemd, upstart, forever, pm2, Docker, etc.
npm install -g process-babysitter
mkdir config
touch config/default.json
# set up config files
process-babysitter ./config
open http://localhost:9876/ # port is configurable, see below
The babysitter uses Node Config (docs) to manage configuration files. This allows multiple configurations to reside in the same directory & extend the base config for various environments & deployments. Some fields can be Lodash templates as noted below.
The configuration should contain:
server
: Server settingsport
: Port to bind to for HTTP & Socket.io access.
template_vars
: Dictionary of variables to make available to process field templates.processes
: Dictionary of processes to babysit. Each processes key should be the same as its id field.id
: Process id. Used as id in HTTP routes.enabled
: Boolean whether the process should be ignored.name
: Human name for process. Displayed in web UI.command
: Templated command to run.cwd
: Templated value for the subprocess working directory.args
: Array of templated arguments for the subprocess (everything that follows the command).env
: Map of environment variable name to value to use for that var. If the value starts with$
, it will be assumed to refer to an existing env var and use its value at run time.
Example default.json
:
{
"server": {
"port": 9876
},
"template_vars": {
"services_dir": "/Users/me/dev/my-stuff/"
},
"processes": {
"api": {
"id": "api",
"enabled": true,
"name": "API Service",
"command": "node",
"cwd": "<%= services_dir %>my-api",
"args": ["<%= services_dir %>my-api/index.js"],
"env": {
"MY_VAR_1": "foo",
"MY_VAR_2": "$BAR"
}
},
"mongodb": {
"id": "mongodb",
"enabled": true,
"name": "MongoDB Server",
"command": "mongod"
}
}
}
It should be noted that child processes will inherit the environment variables set when the babysitter is started. If env vars are specified for the process, they will override the babysitter's env vars for that process.
Use the standard node-config variables (NODE_ENV
, NODE_APP_INSTANCE
, etc) to change which config files are loaded. The babysitter will pass on any process.env
variables except the following so as to not interfere with child processes.
NODE_CONFIG_DIR
NODE_ENV
NODE_APP_INSTANCE
In this manner it is possible to have many project configurations in the same directory (if desired).
The web UI is meant to give an easy interface to start, stop & view process output. Each registered process is listed as a collapsable section and can be started/stopped individually or all at once. The UI interacts with the server via the REST API & Socket.io events. See the source (it's quite small) for the HTTP routes & socket.io events.
The UI itself should be fairly self-explanatory with stdout on the left & stderr on the right. The "pause" button will pause stdout/stderr for all processes (in the UI, the server still collects it) until clicked again. The "clear" button will remove all logs for all processes in the UI & server. Both the UI & server keep the last ~100 lines (may be configurable in the future) of output.
open http://localhost:9876/
npm version x.y.z -m '<release notes>'
npm publish