Reporting Service for Grafana
- About the project
- Getting started
- Setup
- Advanced information
- Roadmap
- Contributing
- Licensing
- Acknowledgements
Grafana is a great tool for monitoring and visualizing data from different sources. Anyway the free version is lacking a professional reporting functionality. And this is, where the ruby grafana reporter steps in.
The key functionality of the reporter is to capture data and images from grafana dashboards and to use it in your custom templates to finally create reports in PDF (default), HTML, or any other format.
By default (an extended version of) Asciidoctor is enabled as template language.
- Create an automated PDF report about your server infrastructure health for your management
- Allow users to build an on-demand CSV file containing data shown on your dashboard, for further use in Excel
- Export your home meter data as a static web-page, that you can publish to the web
- Use Asciidoctor or ERB template language for defining templates
- Render your templates as PDF (default), HTML, CSV, text files and any other format you can think of
- Create reports containing data from multiple grafana dashboards or even multiple grafana installations
- Easy-to-use configuration wizard including automated demo report creation
- Include dynamic content from grafana (find here a reference for all
asciidcotor reporter calls):
- panels as images
- tables based on grafana panel queries or custom database queries (no images!)
- single values to be integrated in text, based on grafana panel queries or custom database queries
- Runs as
- webservice to be called directly from grafana
- standalone command line tool, e.g. to be automated with
cron
orbash
scrips - microservice from standard asciidoctor docker container without any dependencies
- Integrate it in your toolchain with webhooks and API calls
- Solid as a rock - no matter if you do mistakes in your configuration or grafana does no longer match templates: the ruby-grafana-reporter webservice will always return properly.
- Full API documentation available
Functionalities are provided as shown here:
Database | Image rendering | Raw queries | Composed queries |
---|---|---|---|
all SQL based datasources | supported | supported | supported |
Graphite | supported | supported | supported |
InfluxDB | supported | supported | supported |
Prometheus | supported | supported | n/a in grafana |
other datasources | supported | not supported | not supported |
The characteristics of a raw query are, that the query is either specified manually in the panel specification or in the calling template.
Composed queries are all kinds of query, where the grafana UI feature (aka visual editor mode) for query specifications are used. In this case grafana is translating the UI query specification to a raw query, which then in fact is sent to the database.
You don't have a grafana setup runnning already? No worries, just configure
https://play.grafana.org
in the configuration wizard and see the magic
happen!
If your grafana setup requires a login, you'll have to setup an api key for the reporter. Please follow the steps described here first.
Windows:
- Download latest Windows executable
ruby-grafana-reporter -w
Raspberry Pi:
sudo apt-get install ruby
gem install ruby-grafana-reporter
ruby-grafana-reporter -w
Ruby environment:
gem install ruby-grafana-reporter
ruby-grafana-reporter -w
Docker environment (advanced users):
- Download latest single-rb file to an empty folder
- create a configuration file by calling
ruby ruby-grafana-reporter -w
(if in doubt, run the command within your docker container) - create file
/<<path-to-single-rb-file-folder>>/startup.sh
with the following content:
cd /documents
ruby bin/ruby-grafana-reporter
- add the startup script to your asciidoctor section in your docker-compose.yaml:
asciidoctor:
image: asciidoctor/docker-asciidoctor
container_name: asciidoctor
hostname: asciidoctor
volumes:
- /<<path-to-single-rb-file-folder>>:/documents
command:
sh /documents/startup.sh
restart: unless-stopped
- start/restart the asciidoctor docker container
For using the reporter directly from grafana, the reporter has to run as webservice, i.e. it has to be
called without the -t
parameter.
If this is the case, you simply simply need add a link to your grafana dashboard:
- Open the dashboard configuration
- Select
Links
- Select
Add
- Fill out as following:
- Type:
link
- Url:
http://<<your-server-url>>:<<your-webservice-port>>/render?var-template=demo_report
- Title:
Demo Report
- Select
Time range
- Select
Variable values
- Type:
- Select
Add
Now go back to your dashboard and click the newly generated Demo Report
link on it. Now the renderer should start it's task and show you the expected
results.
Please note, that the reporter won't automatically refresh your screen to update
the progress. Simply hit F5
to refresh your browser. After the report has been
successfully built, it will show the PDF after the next refresh automatically.
You want to select a template in grafana, which shall then be rendered?
Piece of cake: Just add a dashboard variable to your grafana dashboard named
template
and let the user select or enter a template name. To make use of it,
you should change the link of the Demo Report
link to
http://<<your-server-url>>:<<your-webservice-port>>/render?
. On
hitting the new link in the dashboard, grafana will add the selected template as
a variable and forward it to the reporter.
It is common practice to use dashboard variables in grafana, to allow users to show the dashboard for a specific set of data only. This is where grafana variables are used.
Those variables are then also used in panel queries, to react on selecting or entering those variables.
You may provide those variables during report generation to the reporter. Therefore you have to specify them in the individual calls.
Let's say, you have a variable called serverid
in the dashboard. You may now want
to set this variable for a panel image rendering. This can be done with the following
calls:
grafana_panel_image:1[var-serverid=main-server]
grafana_panel_image:1[var-serverid=replica-server]
This will render two images: one for main-server
and one for replica-server
.
So, to forward grafana variables to the reporter calls, you simply have to use the
form var-<<your-variable-name>>
and specify those in your reporter template.
Running the reporter as a webservice provides the following URLs
/overview - for all running or retained renderings
/render - for rendering a template, 'var-template' is the only mandatory GET parameter, all parameters will be passed to the report templates as attributes
/view_report - for viewing the status or receving the result of a specific rendering, is automatically called after a successfull /render call
/cancel_report - for cancelling the rendering of a specific report, normally not called manually, but on user interaction in the /view_report or /overview URL
The main endpoint to call for report generation is configured in the previous chapter Grafana integration.
However, if you would like to see, currently running report generations and previously generated reports, you may want to call the endpoint /overview
.
If you want to call the reporter programatically, you might want to get machine usable endpoints. Therefore the following API endpoints are available:
POST /api/v1/render
GET /api/v1/status
DELETE /api/v1/cancel
Those endpoints have to be called with the same URL parameters, as the webservice endpoints.
If you want to receive the report itself, you'll have to use the /view_report
call from the webservice endpoints.
By default the configuration wizard will setup the reporter with the asciidoctor template language enabled. For several reasons, you may want to take advantage of the ruby included ERB template language.
Anyway you should consider, that ERB templates can include harmful code. So make sure, that you will only use ERB templates in a safe environment.
To enable the ERB template language, you need to modify your configuration file
in the section grafana-reporter
:
grafana-reporter:
report-class: GrafanaReporter::ERB::Report
Restart the grafana reporter instance, if running as webservice. That's all.
In ERB templates, you have access to the variables report
, which is a reference
to the currently executed
ERB Report object
and attributes
, which contains a hash
of variables, which have been handed over to the report generations, e.g. from
a webservice call.
To test the configuration, you may want to run the configuration wizard again, which will create an ERB template for you.
Webhooks provide an easy way to get automatically informed about the progress of a report. The nice thing is, that this is completely independent from running the reporter as webservice, i.e. these callbacks are also called if you run the reporter standalone.
To use webhooks, you have to specify, in which progress states of a report you
are interested. Therefore you have to configure it in the grafana-reporter
section of your configuration file, e.g.
grafana-reporter:
callbacks:
all:
- http://<<your_callback_url>>
on_before_create:
- http://<<your_callback_url>>
on_after_cancel:
- http://<<your_callback_url>>
on_after_finish:
- http://<<your_callback_url>>
Remember to restart the reporter, if it is running as a webservice.
After having done so, your callback url will be called for each event with a JSON body including all necessary information of the report. For details see callback.
The reporter is designed to allow easy integration of your own plugins, without having to modify the reporter base source on github (or anywhere else). This section shows how to implement and load a custom datasource.
Implementing a custom datasource is needed, if you use a custom datasource grafana plugin, which is not yet supported by the reporter. In that case you can build your own custom datasource for the reporter and load it on demand with a command line parameter, without having to build your own fork of this project.
This documentation will provide a simple, but mocked implementation of an imagined grafana datasource.
First of all, let's create a new text file, e.g. my_datasource.rb
with the
following content:
class MyDatasource < ::Grafana::AbstractDatasource
def self.handles?(model)
tmp = new(model)
tmp.type == 'my_datasource'
end
def request(query_description)
# see https://rubydoc.info/gems/ruby-grafana-reporter/Grafana/AbstractDatasource#request-instance_method
# for detailed information of given parameters and expected return format
# TODO: call your datasource, e.g. via REST call
# TODO: return the value in the needed format
end
def raw_query_from_panel_model(panel_query_target)
# TODO: extract or build the query from the given grafana panel query target hash
end
def default_variable_format
# TODO, specify the default variable format
# see https://rubydoc.info/gems/ruby-grafana-reporter/Grafana/Variable#value_formatted-instance_method
# for detailed information.
end
end
The only thing left to do now, is to make this datasource known to the
reporter. This can be done with the -r
command line flag, e.g.
ruby-grafana-reporter -r my_datasource.rb
The reporter implemented some magic, to automatically register datasource
implementations on load, if they inherit from ::Grafana::AbstractDatasource
.
This means, that you don't have to do anything else here.
Now the reporter knows about your datasource implementation and will use it,
if you request information from a panel, which is linked to the type
my_datasource
as specified in the handles?
method above. If any errors
occur during execution, the reporter will catch them and show them in the error
log.
Registering a custom ruby file is independent from running the reporter as a webservice or as a standalone executable. In any case the reporter will apply the file.
Technically, loading your own plugin will call require for your ruby file, after all reporter files have been loaded and before the execution of the webservice or a rendering process starts.
This is just a collection of things, I am heading for in future, without a schedule.
- Support grafana internal datasources
- Support additional templating variable types
- Solve code TODOs
- Become rubocop ready
If you'd like to contribute, please fork the repository and use a feature branch. Pull requests are warmly welcome.
If you enjoy my work, and can't support with code contributions, feel free to
Thanks for your support and keeping this project alive!
The code in this project is licensed under MIT license.
Inspired by Izak Marai's grafana reporter