Skip to main content

Best practices for plugin development

We have compiled a list of practices for building and publishing high-quality plugins for Grafana that we have found beneficial. Refer to them when building your own plugin for best results.

Is something missing from this list? Let us know.

General​

  • Verify that your data source or app plugin can be provisioned - Refer to Provisioning for more information.
  • Include default dashboards in your data source or app plugin - Refer to Bundling of dashboards for more information.
  • Ensure that the minimum version of Grafana is correct - Make sure that the grafanaDependency in your plugin.json points to the earliest version of Grafana that your plugin fully supports.
  • Don't expose sensitive information - For security reasons, avoid exposing sensitive information such as secrets. Make sure to utilize log levels properly, avoid excessive logging, and never log credentials or other sensitive information.
  • Avoid using console.log in your plugin - Console messages are usually for debugging purposes and therefore not suitable to ship to the client.
  • Add linting and auto-completion - Reduce bugs in your plugin by adding a code snippet like this one in VS Code to get linting for your plugin.
  • Include a well-written README - Give users a deeper understanding of how to configure and use your plugin, but don’t make it essential reading. You want users to be able to understand your plugin intuitively without referring to the documentation if possible.
  • Allow incremental learning - Hide advanced options using switches or categories, and let users learn about advanced features when they’re ready.
  • Get beta testers - Enlist users in your target audience to try out your plugin before you submit it. Get feedback to help improve your plugin before it's published.

Panel plugins​

  • Don't store or use credentials - Panel plugins don't have a way to securely store credentials. If your plugin needs to use credentials, consider using a data source or app plugin instead and a panel plugin to display the information returned by the data source.

  • Consider creating custom options - If the default panel options aren't a good fit for what you're trying to offer users, use custom options.

  • Document the dataframe schema - Consider documenting the plugin's schema (expected fields, field types, naming conventions for field names, etc.) in the README.

Data source plugins​

  • If your plugin needs to store credentials, use secureJsonData instead of jsonData - The former is encrypted at rest while the latter isn't. Refer to Secure JSON data for more information.
  • Implement a query builder - This is highly useful for users who are not familiar with the query language of the data source. Refer, for example, to the Query Builder for Microsoft SQL Server which helps write SQL queries for that service.
  • Add a health check for your plugin - Health checks are used to verify that the data source is working properly. How this is implemented depends on whether the plugin has a backend. Refer to our examples for health checks in the frontend and backend. For the backend case, it's not necessary to modify its frontend code as long as it extends the DataSourceWithBackend class from @grafana/runtime.
  • Add dashboard variables support - Dashboard (or template) variables allow users to create dynamic dashboards. There are two aspects of adding support for variables in your plugin. The first is allowing queries for your data source and return values to be used as variables. The second is replacing existing variables in other queries. For more information, refer to our documentation. Pay special attention when selecting "All values" since it may require specific logic to join variable values.
  • Add annotations support - Annotations allow users to add contextual information to their dashboards and it's possible to use queries to define them. For more information, refer to Enable annotations.
  • Practice good front-end design - When building frontend components, make sure to use Grafana components as the base and follow the Saga Design System.
  • Add query editor help - Query editors can be complex and it's useful to provide help to users. For more information, refer to Add query editor help.
  • Skip hidden or empty queries - This avoids executing unnecessary or wrong queries. Data sources implementing DataSourceWithBackend only need to implement the method filterQuery. Refer to this example.
  • Specify a default query - Default queries can help users to discover how queries are written for the plugin. Refer to this example.

Frontend (only) plugins​

  • Data sources running only on the frontend typically use the Grafana proxy to access an external service - This is a simple way of adding support for queries in your plugin, and it doesn't require Golang knowledge. However, there are use cases for which writing a backend plugin is necessary. Refer to Backend plugins for more information about those.

Backend plugins​

  • Add support for alerting - Backend plugins have inherent support for Grafana Alerting but this support needs to be enabled. Simply add "alerting": true to your plugin.json file.
  • Use the CallResourceHandler interface to serve custom HTTP requests - For more information, refer to Resource handlers. This is useful, for example, when providing query builders, as shown in this example.
  • Add logs, metrics and traces to your data source. Make it easier to diagnose and resolve issues for both plugin developers and Grafana operators. Find more information in our documentation.
  • Keep cached connections - This is an important optimization. To learn more, refer to our documentation and an example.
  • Add macro support - Macros are similar to variables, but they are typically evaluated in the backend and can return values based on environment data like the current time selection. It can be useful, for example, to evaluate alerts during a dynamic period. These are usually defined using the syntax $__macroName (for example, $__timeFilter). Refer to this example to discover how you can implement support. Some pre-defined macros are available in the plugin SDK macros package.
  • For SQL data sources, refer to the sqlutil package in the SDK - It includes multiple helpers to work with SQL data sources for things like data frame conversion, default macros, and so on. Also, consider using the sqlds package which highly simplifies the implementation of SQL data sources.
  • Don't use the local file system - Different plugins share the same environment. For security reasons, plugins shouldn't rely on local files.
  • Don't use environment variables - Environment variables are also a security risk and should be avoided. For configuration to a particular data source, use the jsonData or secureJsonData fields in the plugin.json file. If configuration is needed for the plugin which is shared among data sources, then use the plugin configuration.
  • Plugins should not execute arbitrary code in the backend - Again, this is a security risk and should be avoided. If your plugin needs to execute code, provide a list of allowed commands and validate the input before executing it.
  • In general, any error happening should be logged with level error.
  • Don't use the info level: use the debug level instead.

App plugins​

  • Specify a root page for your app - If your app defines multiple pages, make sure to select a default one that will be used as a landing page for your plugin.
  • To generate dynamic apps, consider using Grafana Scenes.
  • Consider contributing a UI extension - UI extensions can help a user to discover your app in context and continue a given workflow. Additionally, if your app provides context that can be used in other apps, then create an extension point to allow these apps to do so, with no further changes required in your app.

Publishing a plugin​

  • Add a GitHub badge - Follow these steps to help users find your plugin using GitHub badges.
  • Add workflow automation - If your plugin is available on GitHub, consider adding the GitHub workflows for plugin development to your repository.