Metabase is a widely-used open-source business intelligence tool that aims to simplify data analysis and visualization for all. Its appeal lies in its user-friendly interface and open-source nature, but when it comes to automating workflows and integrating it into complex systems, significant challenges emerge that can make it a less-than-ideal choice for certain use cases. In this lightning talk, I’ll share my firsthand experiences and the hurdles I encountered while attempting to automate our setup with Metabase. These challenges include:
Inadequate API Documentation: While Metabase’s internal documentation covers many features, it falls short in explaining specific parameters in API payloads. Some parameters can be inferred through common sense, but undocumented or unclear ones leave developers uncertain about the correctness of their implementations, hindering effective automation.
Infeasible API-Based Model Conversion: Converting questions to models is straightforward with a single click in the UI, but replicating this via the API is nearly impossible due to excessively large and complex payloads. The API is not designed for external use, rendering this automation task impractical.
Permissions Payload Disconnect: The UI generates lengthy payloads for permission updates, which developers might try to replicate by inspecting network calls in developer tools. However, the backend supports more targeted updates, creating a confusing disconnect that complicates automation efforts.
Visualization Constraints: Basic tasks, like hiding columns in a visualization, require creating new tables or views, adding unnecessary ETL overhead.
jsonb Column Struggles: The query builder fails to handle jsonb columns efficiently, pushing users toward additional data processing steps.
Report Creation Limits: Certain reports can’t be built with the query builder or API, forcing reliance on raw SQL.
Resource-Heavy Syncing: Metabase’s database syncing process is computationally intensive, straining system resources.
Through these real-world examples, I’ll highlight why Metabase’s limitations can outweigh its benefits for teams prioritizing automation. This talk will help attendees evaluate whether Metabase aligns with their needs or if alternative tools might better serve their goals.
Metabase’s incomplete API documentation and impractical payloads hinder effective automation.
Workarounds for visualization and data processing issues can be time-consuming and resource draining.
Teams should weigh their automation requirements carefully before adopting Metabase.
The lightning talk should be supported by a blog from the submitter. I see that Metabase has a lot of issues but what were the solutions .
The flow of designing the solution is not clear here.
This seems a little vague to be a talk. Metabase is a tool with shortcoming like any tool. But like with any tool you may just be holding it wrong.
Coming up with a concrete alternative proposal of another real or theoretical tool to fill this automation gap sounds like what you are trying ro get at, but you haven't explained what that alternative would be.
+1 with other reviewers. I don't think issues with a tool would qualify as a FOSS talk.
Not relevant to the conference, and doesnt add any value without discussing solutions or experience in fixing issues in the project in the past. The FOSS way is to help fix issues rather than announce them.
The reviewers felt that the talk was too focused on the shortcomings of a tool without offering solutions or discussing your experience in helping to fix these issues. Our conference values proposals that reflect the FOSS ethos of contributing to a project to solve problems, rather than just highlighting them. For future submissions, we encourage you to propose a talk that details a problem you solved with a FOSS project or a contribution you made to its codebase.