How-To: Migrating Databricks workspaces

Foreword:
The approach described in this blog post only uses the Databricks REST API and therefore should work with both, Azure Databricks and also Databricks on AWS!

It recently had to migrate an existing Databricks workspace to a new Azure subscription causing as little interruption as possible and not loosing any valuable content. So I thought a simple Move of the Azure resource would be the easiest thing to do in this case. Unfortunately it turns out that moving an Azure Databricks Service (=workspace) is not supported:

Resource move is not supported for resource types ‘Microsoft.Databricks/workspaces’. (Code: ResourceMoveNotSupported)

I do not know what is/was the problem here but I did not have time to investigate but instead needed to come up with a proper solution in time. So I had a look what needs to be done for a manual export. Basically there are 5 types of content within a Databricks workspace:

  • Workspace items (notebooks and folders)
  • Clusters
  • Jobs
  • Secrets
  • Security (users and groups)

For all of them an appropriate REST API is provided by Databricks to manage and also exports and imports. This was fantastic news for me as I knew I could use my existing PowerShell module DatabricksPS to do all the stuff without having to re-invent the wheel again.
So I basically extended the module and added new Import and Export functions which automatically process all the different content types:

  • Export-DatabricksEnvironment
  • Import-DatabricksEnvironment

They can be further parameterized to only import/export certain artifacts and how to deal with updates to already existing items. The actual output of the export looks like this and of course you can also modify it manually to your needs – all files are in JSON except for the notebooks which are exported as .DBC file by default:

A very simple sample code doing and export and an import into a different environment could look like this:

Having those scripts made the whole migration a very easy task.
In addition, these new cmdlets can also be used in your Continuous Integration/Continuous Delivery (CI/CD) pipelines in Azure DevOps or any other CI/CD tool!

So just download the latest version from the PowerShell gallery and give it a try!

56 Replies to “How-To: Migrating Databricks workspaces”

  1. Can I use your DatabricksPS module to add environmental variables to clusters? Also, can I restrict the export to just workspaces (i.e. no clusters or jobs) using the Export-DatabricksEnvironment?

  2. I have Owner role on my databricks workspace that I want to import and export, and while I try to export the whole workspace, I get the following errors

    MY Powershell version – 7.1.0, running on Windows 10 Enterprise

    Commands Run –
    PS C:\Windows\System32> Set-DatabricksEnvironment -AccessToken -ApiRootUrl “https://westeurope.azuredatabricks.net”
    PS C:\Windows\System32> Export-DatabricksEnvironment -LocalPath ‘C:\Databricks\Export’ -CleanLocalPath
    WARNING: This feature is EXPERIMENTAL and still UNDER DEVELOPMENT!
    WARNING: LIBRARY found at /Users//spark-xml_2.12-0.9.0 – Exporting Libraries is currently not supported!
    WARNING: It is not possible to extract secret values via the Databricks REST API.
    This export only exports the names of SecretScopes and their Secrets but not the values!
    WARNING: It is not possible to donwload the whole DBFS.
    This export will only download files from DBFS that already exist locally and overwrite them!
    Export-DatabricksEnvironment: Local DBFS path C:\Databricks\Export\DBFS does not exist so the DBFS export cannot work properly!

    How do I import all the artefacts, for a complete workspace migration?

    • Correction: I noticed, the content put in angle brackets are missed out. So reposting a warning message regarding libraries again, from above.


      WARNING: LIBRARY found at /Users/another-user/spark-xml_2.12-0.9.0 – Exporting Libraries is currently not supported!

        • How do I resolve the below Warning?

          WARNING: It is not possible to donwload the whole DBFS.
          This export will only download files from DBFS that already exist locally and overwrite them!
          Export-DatabricksEnvironment: Local DBFS path C:\Databricks\Export\DBFS does not exist so the DBFS export cannot work properly!

          Right now I’m performing this from my local desktop.

          Also, a small correction to your logging module, the spelling for download is incorrect ?

          • You can use the parameter “-Artifacts” and provide e.g. Only Workspace, Clusters and Jobs as array so DBFS and Secrets would not be exported and you also would not get a warning

  3. Can I export the notebooks as a python / SQL file?
    My plan is to commit the code to Azure repo and sync it with azure pipeline

    • yes, you can definitely do that
      take a look at the

      and

      cmdlet described in the blog post.
      they allow you to export and import various items like the workspace (which includes the notebooks) via PowerShell which can then be triggered via Azure DevOps for example

      regards
      -gerhard

    • the

      cmdlet can also be used to import DBFS items
      for this to work you need to place the files you want to upload in a subfolder called “DBFS” under your -LocalPath parameter

      Assuming you have your files under c:\Content\DBFS you can use the following code snippet to upload them to DBFS:

        • the

          works slightly different as we obviously cannot just download the whole DBFS it will just download the files that already exist in the local folder

        • currently the export Export-DatabricksEnvironment takes the local files in your output-DBFS folder and downloads them again
          so if you have \DBFS\myfile.txt it will download myfile.txt from the root of your DBFS when you run Export-DatabricksEnvironment
          so basically the locally existing files control what is downloaded
          you can also create an empty local file to force a download

          This is by design as it does not make sense to download the whole DBFS

          • @gerhard brueckl I have some DBFS file locally, but after running this command no output related to DBFS appears

              • Export-DatabricksEnvironment -Artifacts \DBFS\temp\LoanStats3a.csv -LocalPath ‘D:\MyExport’

                • I think what you are looking for is Download-DatabricksFSFile
                  also, your syntax for Export-DatabricksEnvironment would be wrong, it should be
                  you would need to use Export-DatabricksEnvironment -Artifacts DBFS -LocalPath ‘D:\MyExport’

                    • it would be really helpful if you provide any more information than that, its pretty hard to debug otherwise …
                      so whats the the PS cmd you executed?
                      do you get any error messages?
                      have you tried using the -Verbose flag?
                      do the other cmdlets work?
                      does the DBFS file exist?

    • to be honest, I am not aware of a good solution for databases/SQL objects
      you cannot import/Export them as there are no APIs – the only way would be to execute some code to get the definition of the SQL objects
      but I have not automated something like this yet

  4. how to move my existing DataBricks notebooks to azure synapse workspace ?
    any help would be much appreciated.

  5. i am not able to import the users. its only importing the groups & not adding the members to it. giving me the below error.

    Invoke-RestMethod : {“error_code”:”RESOURCE_DOES_NOT_EXIST”,”message”:”User
    ‘surbhi.bhan@abc.com’ does not exist”}
    At C:\Program Files\WindowsPowerShell\Modules\DatabricksPS\1.9.6.1\Public\General.ps1:87 char:13
    + … $result = Invoke-RestMethod -Uri $apiUrl -Method $Method -Headers $ …
    + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CategoryInfo : InvalidOperation: (System.Net.HttpWebRequest:HttpWebRequest) [Invoke-
    RestMethod], WebException
    + FullyQualifiedErrorId : WebCmdletWebResponseException,Microsoft.PowerShell.Commands.InvokeRes
    tMethodCommand

    • well, the error message basically tells you whats wrong: ”User ‘surbhi.bhan@abc.com’ does not exist”
      maybe you have moved a workspace from one tenant to another where this user has not been added yet?

      -gerhard

    • Hi Praveen,
      no, there is no built-in way to do this. However, you can use a combination of

      and

      to download DBFS content recursively

      -gerhard

  6. Hi,

    I’ve tried to use the Import-DatabricksEnvironment cmdlet in an Azure DevOps pipeline and it gives me the following error:

    Cannot process command because of one or more missing mandatory parameters: NodeTypeId SparkVersion.

    I specifically done this:

    Import-DatabricksEnvironment
    -LocalPath "$(Agent.BuildDirectory)/s/[FOLDER]"

    When I ran this locally, I had to provide the values for NodeTypeId and SparkVersion via the prompt after executing the above cmdlet.

    I executed the cmdlet locally like this:

    Import-DatabricksEnvironment
    -Artifacts Workspace,Clusters,Jobs

    -LocalPath “[LOCAL DIRECTORY]”

    Do you have any ideas on how I can execute Import-DatabricksEnvironment in an Azure DevOps pipeline?

    Many Thanks!

    • this must be related to the Clusters being imported – so I guess in your cluster definitions the NodeTypeID and SparkVersion is missing – can you check for this?
      can you share the definition of the cluster(s)?

      • My cluster definition is this:

        {
        “autoscale”: {
        “min_workers”: 2,
        “max_workers”: 8
        },
        “cluster_name”: “Test”,
        “spark_version”: “9.1.x-scala2.12”,
        “spark_conf”: {},
        “azure_attributes”: {
        “first_on_demand”: 1,
        “availability”: “ON_DEMAND_AZURE”,
        “spot_bid_max_price”: -1
        },
        “node_type_id”: “Standard_DS3_v2”,
        “driver_node_type_id”: “Standard_DS3_v2”,
        “ssh_public_keys”: [],
        “custom_tags”: {},
        “spark_env_vars”: {
        “PYSPARK_PYTHON”: “/databricks/python3/bin/python3”
        },
        “autotermination_minutes”: 120,
        “enable_elastic_disk”: true,
        “cluster_source”: “UI”,
        “init_scripts”: [],
        “cluster_id”: “0106-173902-81dw5o61”
        }

        It’s just a blank cluster I created for the purpose of testing the import and export of the Databricks’ contents from one resource to another.

        So when I did run this locally like mentioned in my first message, the process did work and I could see the cluster in the other resource.

        So would there be a way for me to provide values for the nodeTypeId and sparkVersion in an Azure DevOps pipeline? I did try this and got the following error (since there aren’t parameters for the nodeTypeId/sparkVersion):

        A parameter cannot be found that matches parameter name ‘NodeTypeId’.

        FYI the PowerShell used:

        Set-DatabricksEnvironment
        -AccessToken [dbw-001-token]

        -ApiRootUrl [dbw-001-URL]

        Export-DatabricksEnvironment
        -LocalPath '$(Agent.BuildDirectory)/s/[DIRECTORY]'

        -CleanLocalPath
        -Artifacts Workspace,Clusters,Jobs

        Set-DatabricksEnvironment
        -AccessToken [dbw-002-token]
        -ApiRootUrl [dbw-002-URL]

        Import-DatabricksEnvironment
        -LocalPath “$(Agent.BuildDirectory)/s/[DIRECTORY]”
        -NodeTypeId "Standard_DS3_v2"

        -SparkVersion “9.1.x-scala2.12”

        Let me know your thoughts and thank you for your reply 🙂

  7. I have used this before but I obviously doing something wrong.

    I have created a Access Token in Databricks, connected to Azure in Powershell using Connect-AzAccount and set the token.

    However I am getting this when trying to export

    Export-DatabricksEnvironment -LocalPath ‘C:\Testing\MyExport’ -CleanLocalPath
    WARNING: This feature is EXPERIMENTAL and still UNDER DEVELOPMENT!
    Invoke-RestMethod :
    Error 403 Invalid access token.
    HTTP ERROR 403
    Problem accessing /api/2.0/workspace/list. Reason:
    Invalid access token.

    Is there a specific way to export the token that I am missing?

  8. Hi,
    This is great. Thanks for sharing! I was able to successfully import the cluster and workspace. However, I get an error when I try to import jobs. Please see the call stack below. Any suggestions on how to resolve the issue? I would appreciate it!

    InvalidOperation: \PowerShell\Modules\DatabricksPS\1.11.0.8\Private\General.ps1:396:5
    Line |
    396 | … $hash[$object.Key] = ConvertTo-Hashtable $object.Value
    | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    | Index operation failed; the array index evaluated to null.
    Invoke-RestMethod: PowerShell\Modules\DatabricksPS\1.11.0.8\Public\General.ps1:96:13
    Line |
    96 | … $result = Invoke-RestMethod -Uri $apiUrl -Method $Method -Headers $ …
    | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    | {“error_code”:”MALFORMED_REQUEST”,”message”:”Could not parse request object: Expected Scalar value for String field ‘on_failure’\n at
    | [Source: (ByteArrayInputStream); line: 1, column: 458]\n at [Source: java.io.ByteArrayInputStream@248fbab6; line: 1, column: 458]”}
    Invoke-RestMethod: \PowerShell\Modules\DatabricksPS\1.11.0.8\Public\General.ps1:96:13
    Line |
    96 | … $result = Invoke-RestMethod -Uri $apiUrl -Method $Method -Headers $ …
    | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    | {“error_code”:”INVALID_PARAMETER_VALUE”,”message”:”One of job_cluster_key, new_cluster, or existing_cluster_id must be specified.”}

  9. We have new envt in Databricks where the repository is cloned to Azure DevOps. Wnen we do the bulk import from lower envt, is there a way to bulk import into Databricks repository also?

    • if you already have your code in a repo, why would you import it again into your new workspace and not simply attach the repo in the new workspace too?

  10. I’m running the latest version of DatabricksPS, I have admin on my Databricks workspace, and Owner on my Azure subscription. I can run set-databricksenvironment with my access token and API root URL for East US, and the script appears to be able to view the users and folders, but when I run Export-databricksenvironment -localpath “c:\dbricksexport” -cleanlocalpath it throws the same error continuously as it runs:

    Exception calling “WriteAllBytes” with “2” argument(s): “The given path’s format is not supported.”
    At C:\Users\myusername\OneDrive – mycompanyname\Documents\WindowsPowerShell\Modules\DatabricksPS\1.12.0.1\Public\WorkspaceAPI.ps1:91 char:3
    + [IO.File]::WriteAllBytes($LocalPath, $exportBytes)
    + ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    + CategoryInfo : NotSpecified: (:) [], MethodInvocationException
    + FullyQualifiedErrorId : NotSupportedException

    Anything I can attempt to fix?

Leave a Reply to Gerhard Brueckl Cancel reply

Your email address will not be published. Required fields are marked *

*