About

This is a prototype to do deep copies of Metabase collections.

Requirements

• A Metabase instance with a Pro license, to unlock the serialization feature to import/export the collection
• An API key, which can be created on Metabase at /admin/settings/authentication/api-keys
• A collection you want to clone or change the data source for

Installation

• Make sure you have Ruby installed
• Check out this repo with git clone https://github.com/metabase/template-clone-prototype.git
• Install the dependencies with bundle install
• Test if the dependencies are satisfied by running ruby clone.rb --help.

Supported use cases and usage

1. Clone a collection, optionally updating the underlying data source
2. Change the data source of a collection without duplicating or renaming it

Limitations

It can only swap out one data source at a time. If the items in your collections depend on multiple data sources, you might have to do several runs. This scenario has not been tested yet.

Example usage

1. Locate the ID of the collection to use as template. In this example it's 9. Export the collection via serialization API by specifying collection=ID.

curl \        -H 'x-api-key: API_KEY' \  
-X POST 'http://localhost:3000/api/ee/serialization/export?settings=false&data_model=false&collection=9' \                                      
-o metabase_data.tgz

2. Extract data from the tarball. This will create a directory with YAML files and subdirectories inside of it. The directory name is based on the site name and timestamp of the export.

tar xzf metabase_data.tgz

3. Run this tool pointing to the directory extracted from the tarball, and providing the source and target names for the data source names.

If want ato load a copy into the same instance, you will want to use the following parameters:

• the --source-datasource and --target-datasource to swap out the data source (these are required for now, so if you don't want to change the data source just use the same data source name in both)
• the --duplicate parameter if you want a copy, and omit this parameter if you want to overwrite the collection changing it's datasource
• the --new-collection-name to give the new copy a different name or to change the name of the collection

ruby clone.rb --yaml-files foobar-2024-08-09_20-57 --source-datasource "Old database" --target-datasource "New database" --new-collection-name "Super fancy new collection" --duplicate

• --yaml-files required, should point to the directory you extracted from the tarball
• --source-datasource required, the display name of the data source to replace (check /admin/databases on Metabase)
• --target-datasource required, the display name of the data source to replace with (check /admin/databases on Metabase)
• --new-collection-name optional, the new name for the collection
• --duplicate optional, will create new entity IDs for all items to clone entire collection and its contents

4. Create a tarball with the modified files

tar -czf metabase_data_modified.tgz foobar-2024-08-09_20-57

5. Import the tarball via serialization API

curl -X POST \  -H 'x-api-key: API_KEY' \
  -F file=@metabase_data_modified.tgz \
  'http://localhost:3000/api/ee/serialization/import' \
  -o -

Use cases and parameters

Change the datasource from prod to dev

ruby clone.rb --yaml-files foobar-2024-08-09_20-57 --source-datasource "Production" --target-datasource "Development" --new-collection-name "Collection now pointing to dev"

Clone the collection without changing the datasource

ruby clone.rb --yaml-files foobar-2024-08-09_20-57 --source-datasource "Production" --target-datasource "Production" --new-collection-name "Customer collection copy" --duplicate

Clone the collection, changing the datasource

ruby clone.rb --yaml-files foobar-2024-08-09_20-57 --source-datasource "Production" --target-datasource "Customer A Production" --new-collection-name "Customer A collection" --duplicate