Tool to help you manage your Grafana dashboards using Git.
Brendan Abolivier f2f08e71c9
Don't panic on remote repository is empty
преди 7 години
src Don't panic on remote repository is empty преди 7 години
vendor Add logging преди 7 години
.editorconfig Add .editorconfig преди 7 години
.gitignore Add config file example and ignore the real config file преди 7 години
LICENSE Add AGPL license преди 7 години
README.md Typo преди 7 години
config.example.yaml Add ignored prefix to the configurable keys and filter by dashboard name instead of slug преди 7 години

README.md

Grafana Dashboard Manager

The Grafana Dashboard Manager is a tool to help you manage your Grafana dashboards using Git.

Manager

The manager is split in two parts:

The puller

The puller is a tool that will pull all the dashboards from the Grafana API, except the ones with a name starting with a specific prefix (if provided in the configuration file), and commit them to the Git repository if needed (and push them to the remote afterwards).

To determine if a dashboard sould be commited to the repository, the puller relies on Grafana's dashboard version management. It will store the versions of all known dashboards (in a file called versions.json, which it will create if it doesn't exist), and commit changes to a dashboard only if the version retrieved from the Grafana API has a greater version number than the one stored in versions.json (if none is stored, it will systematically commit the retrieved dashboard).

If a dashboard has changes to be commited, its JSON description will be stored in a JSON file at the root of the repository (named [dashboard slug].json), and will be added to the Git index. Once all new or modified files have been added to the index, the puller creates a commit with the detail of the update in the commit message, then pushes it to the remote.

If there wasn't any error causing it to panic, the puller exits once all commited changes have been pushed to the Git remote.

The pusher

The pusher is a tool that will expose a webhook to a given address (interface:port/path) and process incoming push events sent by the Git remote.

For every push event on the master branch of the repository, it will look at the files added or modifed by the pushed commits (ignoring the ones with a name starting with a specific prefix (if provided in the configuration file)). It will then proceed to push them to the Grafana API to update modified dashboards or create added ones.

It will then call the puller to have all the files up to date. This is mainly done to update the version number of each dashboard, as Grafana updates them automatically when a new or updated dashboard is pushed.

Please note that the pusher currently only pushes new or modified dashboards to the Grafana API. If the file for a dashboard is removed from the Git repository, the dashboard won't be deleted on the Grafana instance.

Because it hosts a webserver, the pusher runs as a daemon and never exists unless it panics because of an error, or it is killed (e.g. with Ctrl+C).

Build

The manager can be built using gb, which can be installed by running

go get github.com/constabulary/gb/...

It can then be built by cloning this repository and running

cd grafana-dashboard-manager
gb build

Once built, binaries are located in the bin directory (which is created by gb if it doesn't exist).

Run

To run either the puller or the pusher, simply execute the corresponding binary

./puller

or

./pusher

Of course, this command line call may depend on the location and name of the binaries.

You can specify a configuration file via the command line flag --config, which works with both the puller and pusher. For example, here's how the full call should look like when passing a configuration file path to the puller:

./puller --config /etc/grafana-dashboard-manager/config.yaml

If the --config flag isn't present in the command line call, it will default to a config.yaml file located in the directory from where the call is made.

Configure

To run either the puller or the pusher, you will need a configuration file first. The simplest way to create one is to copy the config.example.yaml file at the root of this repository and replace all of the placeholder values with your own.

Since all the keys are documented as comments in the config.example.yaml file, there won't be any more documentation about them in this README file.