Deployment

A website is only a website if other people can look at it. While you are developing locally that's not really all that helpful. So how do you get your changes up to your favorite web host? This is where lektor deploy comes in.

Deploying in Two Steps:

Deploying a website in Lektor is a two step process:

  1. build
  2. deploy

Keep this in mind. A deploy will not implicitly build! This means that if you deploy without building first you might send up a completely wrong version! Also more importantly: never deploy unless the build finished successfully.

The Build Pipeline

So let's cover the building first. When you use the Lektor server locally, Lektor constantly builds out your website into static HTML files behind the scenes into the default build folder. This folder is in an operating system specific location. If you want to know where that folder is, you can use this command:

$ lektor project-info --output-path

Additionally you can manually provide a different path if you kick off a manual build:

$ lektor build --output-path my-folder

Generally we strongly recommend to use the default build folder when deploying from your own machine because it will be faster since the build can reuse what the development server already did. If you want to deploy from a build server it might make more sense to provide absolute paths.

Lektor Assisted Deployments

Currently Lektor can deploy via rsync and ftp automatically. To enable this functionality you need to configure this in the config file. For each potential deployment target add a [servers.NAME] section. The supported keys are name for an optional human readable name of the server, enabled to enable or disable it (defaults to true) and target which is the URL to publish to. Additionally one of the servers can have default set to yes to set it as default. Here is an example:

[servers.production]
name = Production
enabled = yes
default = yes
target = rsync://server/path/to/folder

To trigger a deploy you can use the deploy command:

$ lektor deploy production

Or because it's the default server, you can just do this:

$ lektor deploy

This deploys the latest state of what was built from the default build folder. It does not build itself! So to do both in one go, do something like this:

$ lektor build && lektor deploy

If you want to provide a different build folder, provide it explicitly with --output-path to both build and deploy. This can also be set with the LEKTOR_OUTPUT_PATH environment variable, or provided in the project file.

Note on credentials: For FTP and other systems it is possible to embed the credentials directly in the URL. If you do this, please ensure that you keep your project file secure as loss of the project file can mean that people get access to your server. Alternatively you can also provide username and password through the command line or environment variables.

The following targets are supported for the target field natively:

In addition, there are third-party plugins available for additional deploy targets.

Manual Deployments

If you want to manually deploy something through the favorite tool of yours you can do that easily as well. For instance if you want to deploy to S3 with s3cmd you could do this:

$ lektor build && s3cmd sync "$(lektor project-info --output-path)"/* "s3://my-bucket/some/path"

Assisted deployments are also supported directly from the admin UI. There is a publish button that can be used to send the changes up.

Comments