specit is a fork of speculate with some different goals. specit meant to support more distributions, provide even more options. Feel free to create issues, submit more code. Note: Currenly specit doesn't generate systemd services, we plan to return this feature and provide initd template as well for ancient linux distributions.
Automatically generates an RPM Spec file for your Node.js project
npm install --global specit
- Generates an RPM Spec file for your project
- (optional) Creates a systemd service definition file
- Supports configuration using your existing
package.json
Let's start with a simple Node.js project:
my-cool-api
├── node_modules
├── package.json
└── server.js
1 directory, 2 files
Run the specit
command from inside the project directory:
specit
You've now got an RPM Spec file and a systemd service definition for your project. You'll also notice that your application has been packaged into a tar.gz
archive, ready to be built with an RPM building tool like rpmbuild
or mock
:
my-cool-api
├── SOURCES
│ └── my-cool-api.tar.gz
├── SPECS
│ └── my-cool-api.spec
├── node_modules
├── my-cool-api.service
├── package.json
└── server.js
3 directories, 5 files
Specit is designed to be used at build time, just before you package your application into an RPM. Because of this, we recommend adding the generated files to your .gitignore
file:
SOURCES
SPECS
Specit assumes that you've already installed your npm dependencies when it is run. This means that you don't need to worry about running npm install
inside a clean RPM-building environment like mock.
By default specit will set the RPM release number to 1, if you want to override this you can do so by using the --release
flag:
specit --release=7
By default specit will set the name from package.json
, if you want to override this you can do so by using the --name
flag:
specit --name=my-cool-api
This is useful if you are using private NPM packages which start with an @
.
You can then run npm run spec
to generate your spec file in an environment where specit isn't installed globally (like your CI server.)
If you want to use a different specfile template to create your package, you can specify it your package.json
:
{
"spec": {
"specTemplate": "templates/myspec.mustache"
}
}
By default Specit will build noarch
packages meaning the final package should be installable on every CPU Architecture your system runs on.
Specit's default template will also instruct rpmbuild to skip binary stripping during build.
If your Nodejs application has binary modules you may want to disable this behavior through your package.json
.
{
"spec": {
"noarch": false
}
}
To minimise the final RPM size, your development dependencies (dependencies added with the --save-dev flag) are automatically pruned so that they're not shipped with your production code.
If for some reason you need to package your dev dependencies with your production code you can explicity tell specit not to prune by adding the following to your package.json
:
{
"spec": {
"prune": false
}
}
The systemd service file that Specit generates uses the npm start
script to start your application. Make sure that you've defined this script in your package.json
file.
{
"scripts": {
"start": "node server.js"
}
}
By default, the spec file that specit generates isn't tied to a particular Node version. It simply requires the nodejs
package. It's up to you to make the package available when you install the RPM using yum
.
We strongly recommend that you use the Nodesource binary distributions to install a modern version of Node.js for both your RPM building environment and your target server. Follow the setup instructions for Enterprise Linux and then run yum install nodejs
.
If you're using multiple node repositories or a repository with multiple versions of node, you can specify an RPM version requirement with the engines
property in your package.json
file:
{
"engines": {
"node": "< 5.0.0"
}
}
The engines.node
property must conform to the RPM version syntax
Specit creates the following directories for your application:
Directory | Purpose |
---|---|
/opt/:projectName |
This is where your application is stored |
/var/log/:projectName |
This is created for any log files that your application needs to write to |
You can set Installation path by setting the installDir
inside your package.json
:
{
"spec": {
"installDir": "/usr/local"
}
}
To add a dependency to the generated spec file, list the package dependencies in the requires
array:
{
"spec": {
"requires": [
"vim",
"screen"
]
}
}
If you have any build dependencies (such as python
for node-gyp
), instead of having them available outside the build environment you can instead add them to the buildRequires
array:
{
"spec": {
"buildRequires": [
"python"
]
}
}
If you need to perform any actions after installing your package (such as moving files on the target server) you can specify these inline using the post
property:
{
"spec": {
"build": [
"gulp clean",
"gulp build"
]
}
}
If you have scripts that need to be executable when they're installed on your target server, add them to the executable
array. You can list both files and entire directories:
{
"spec": {
"executable": [
"./other-scripts/my-script.js",
"./scripts"
]
}
}
If you need to perform any actions after installing your package (such as moving files on the target server) you can specify these inline using the post
property:
{
"spec": {
"post": [
"mv /opt/my-cool-api/rc.local /etc/rc.local"
]
}
}
If you need to specify environment variables during startup (NODE_ENV for example) you can specify these inline using the spec.environment property:
{
"spec": {
"environment": {
"NODE_ENV": "production",
"NODE_INSTANCE": "%i"
}
}
}
speculate had syslog enabled by default for their generated services, which might cause double logging because systemd already has journald for it's logging purposes. you can enable syslog back if you find it important.
{
"spec": {
"syslog": true
}
}