NOTE: For a more extensive and maintained cross-workload migration solution, please use the Databricks Terraform Exporter, which creates Infrastructure-as-a-Code replicas for the entire manually-configured Databricks Workspaces.
This is a migration package to log all Databricks resources for backup and/or migrating to another Databricks workspace. Migration allows a Databricks organization to move resources between Databricks Workspaces, to move between different cloud providers, or to move to different regions / accounts.
This package is based on python 3.6 and DBR 6.x+ releases.
Python 3.7 or above is recommended if one is also exporting/importing MLflow objects.
Note: This tool does not support windows currently since path resolution is different from mac / linux.
This package also uses credentials from the Databricks CLI.
To use this migration tool, you'll need:
- An environment running linux with python, pip, git, and the databricks CLI installed.
- Admin access to both the old and new databricks accounts in the form of a Personal Access Token.
Click to expand & collapse tasks
1. Generate Tokens
Generate Access Tokens for both the old and new databricks accounts
- Click
User Settings Icon Settings in the lower left corner of your Databricks workspace
- Click on Access Tokens tab
- Click on Generate New Token button.
- Copy the generated token and store in a secure location.
Be sure to keep a file with the url for both the old and new databricks account Add the old and new token and the old and new Instance ID if applicable. You'll need easy access to all of these things when running the migration tool.
2. Setup databricks-cli profiles
In order to run the migration tool from your linux shell
Create a profile for the old workspace by typing:
databricks configure --token --profile oldWSIn this case oldWS is the profile name you'll refer to for running the migration tool export_db.py file within the old databricks account.
When you use the databricks cli configure command, you'll be prompted for 2 things
Databricks Host (should begin with https://): When this happens, enter the old databricks workspace URL that you captured in your file above.Token: When this happens, paste in the token you generated for the old databricks account.
Repeat the steps above for the new databricks account and change the oldWS profile name to something like newWS in order to keep track of which account you're exporting FROM and which account you're importing TO.
Create a profile for the New workspace by typing:
databricks configure --token --profile newWSIn this case newWS is the profile name you'll refer to for running the migration tool import_db.py file within the new databricks account.
3. Install package dependencies
In order to set up the python environment, clone this repository and python3 setup.py install from the top-level project directory.
To use the migration tool see the details below to start running the tool in the order recommended to properly migrate files.
Support Matrix for Import and Export Operations:
| Component | Export | Import |
|---|---|---|
| Users / Groups | Supported | Supported |
| Clusters (w/ ACLs) | Supported | Supported |
| Notebooks (w/ ACLs) | Supported | Supported |
| Repos (w/ ACLs) | Supported | Supported* |
| Metastore | Supported | Supported |
| Jobs (w/ ACLs) | Supported | Supported |
| Libraries | Supported | Unsupported |
| Secrets | Supported | Supported |
| Table ACLs | Supported | Supported |
| DBFS Mounts | Supported | Unsupported |
| ML Models | Supported* | Supported* |
Note on MLFlow Migration:
MLFlow asset migration is currently only partially supported; Feature Store and Model Registry will not be migrated, for example. See mlflow-export-import for comprehensive MLflow migrations.
Note on DBFS Data Migration:
DBFS is a protected object storage location on AWS and Azure. Please contact your Databricks support team for information about migrating DBFS resources.
Note on User Migration:
During user / group import, users will be notified of the new workspace and account by default. To disable this behavior, please contact your Databricks account team.
Note on Repos: Private repos cannot be imported. These should be added manually using the original user credentials.
The recommended method of exporting and importing is by using the Pipeline contained in migration_pipeline.py. This pipeline performs all export and import steps sequentially, and includes checkpointing parallelization features.
python migration_pipeline.py -h
usage: migration_pipeline.py [-h] [--profile PROFILE] [--azure or gcp] [--silent] [--no-ssl-verification] [--debug] [--set-export-dir SET_EXPORT_DIR]
[--cluster-name CLUSTER_NAME] [--notebook-format {DBC,SOURCE,HTML}] [--overwrite-notebooks] [--archive-missing]
[--repair-metastore-tables] [--metastore-unicode] [--skip-failed] [--session SESSION] [--dry-run] [--export-pipeline] [--import-pipeline]
[--validate-pipeline] [--validate-source-session VALIDATE_SOURCE_SESSION] [--validate-destination-session VALIDATE_DESTINATION_SESSION]
[--use-checkpoint] [--skip-tasks SKIP_TASKS [SKIP_TASKS ...]] [--num-parallel NUM_PARALLEL] [--retry-total RETRY_TOTAL]
[--retry-backoff RETRY_BACKOFF] [--start-date START_DATE] [--hipaa] [--skip-large-nb]
[--exclude-work-item-prefixes EXCLUDE_WORK_ITEM_PREFIXES [EXCLUDE_WORK_ITEM_PREFIXES ...]]
Export user(s) workspace artifacts from Databricks
optional arguments for import/export pipeline:
-h, --help show this help message and exit
--profile PROFILE Profile to parse the credentials
--azure or --gcp Run on Azure or GCP (Default is AWS)
--silent Silent all logging of export operations.
--no-ssl-verification
Set Verify=False when making http requests.
--debug Enable debug logging
--no-prompt Skip interactive prompt/confirmation for workspace import.
--set-export-dir SET_EXPORT_DIR
Set the base directory to export artifacts
--cluster-name CLUSTER_NAME
Cluster name to export the metastore to a specific cluster. Cluster will be started.
--notebook-format {DBC,SOURCE,HTML}
Choose the file format to download the notebooks (default: DBC)
--overwrite-notebooks
Flag to overwrite notebooks to forcefully overwrite during notebook imports
--archive-missing Import all missing users into the top level /Archive/ directory.
--repair-metastore-tables
Repair legacy metastore tables
--hipaa Use HIPAA-compatible cluster profiles
--metastore-unicode log all the metastore table definitions including unicode characters
--skip-failed Skip retries for any failed hive metastore exports.
--skip-missing-users Skip failed principles during ACL import; for missing principles, this will result in open ACLs
--skip-large-nb Skip notebooks that exceed the size limit of 10485760 bytes instead of failing
--session SESSION If set, pipeline resumes from latest checkpoint of given session; Otherwise, pipeline starts from beginning and creates a new session.
--dry-run Dry run the pipeline i.e. will not execute tasks if true.
--export-pipeline Execute all export tasks.
--import-pipeline Execute all import tasks.
--use-checkpoint use checkpointing to restart from previous state
--skip-tasks SKIP_TASK [SKIP_TASK ...]
Space-separated list of tasks to skip from the pipeline. Valid options are:
instance_profiles, users, groups, workspace_item_log, workspace_acls, notebooks, secrets,
clusters, instance_pools, jobs, metastore, metastore_table_acls, mlflow_experiments, mlflow_runs
--keep-tasks KEEP_TASK [KEEP_TASK ...]
Space-separated list of tasks to run from the pipeline. See valid options in --skip-tasks. Overrides skip-tasks.
--num-parallel NUM_PARALLEL
Number of parallel threads to use to export/import
--retry-total RETRY_TOTAL
Total number or retries when making calls to Databricks API
--retry-backoff RETRY_BACKOFF
Backoff factor to apply between retry attempts when making calls to Databricks API
--start-date START_DATE
start-date format: YYYY-MM-DD. If not provided, defaults to past 30 days. Currently, only used for exporting ML runs objects.
--groups-to-keep group [group ...]
List of groups to keep if selectively exporting assets. Only users (and their assets) belonging to these groups will be exported.
options for validation pipeline:
--validate-pipeline Validate exported data between source and destination.
--validate-source-session VALIDATE_SOURCE_SESSION
Session used by exporting source workspace. Only used for --validate-pipeline.
--validate-destination-session VALIDATE_DESTINATION_SESSION
Session used by exporting destination workspace. Only used for --validate-pipeline.
To export a workspace, run:
python3 migration_pipeline.py --profile $SRC_PROFILE --export-pipeline --use-checkpoint [--session $SESSION_ID]
Where $SRC_PROFILE is the Databricks profile for the source workspace, as configured during Setup, and $SESSION_ID is an optional session identifier used for subsequent checkpoint runs. All data is exported to a folder named according to the $SESSION_ID value under the logs folder - “logs/$SESSION_ID”. If $SESSION_ID is not specified, a random value will be generated.
As a starting point, we recommend using the following parameter values:
- retry-total=30
- num-parallel=8
- retry-backoff=1.0
These can be adjusted per your scenario if needed; in general, if API limits are being hit, you can increase retry-backoff, decrease num-parallel, or both.
If script failure occurs, you can safely rerun the same command with --use-checkpoint and --session $SESSION_ID to let the migration pick up from the previous checkpoint and rerun.
If your source and destination workspaces are in different accounts, you will need to update the Instance Profile ARN accordingly during the migration. To do this, run the following command after exporting the workspace assets:
python3 export_db.py --profile $SRC_PROFILE --use-checkpoint --old-account-id $OLD_AWS_ACCT_ID --update-account-id $NEW_AWS_ACCT_ID --set-export-dir $EXPORT_DIR/$SESSION_ID
Where EXPORT_DIR/SESSION_ID is the directory and session ID used by your export job, SRC_PROFILE is the profile used to export the source workspace, OLD_AWS_ACCT_ID is the source AWS account ID, and NEW_AWS_ACCT_ID is the destination AWS account ID. Note that this will only update the ARN in the Instance Profiles; the same instance profiles must still exist in the destination workspace.
To import into a target workspace, run:
python3 migration_pipeline.py --profile $DST_PROFILE --import-pipeline --use-checkpoint [--session $SESSION_ID]
The same recommended parameters as above apply in the import workflow, and similarly, if a failure occurs, --use-checkpoint can be used to rerun from the last checkpoint.
Simple workspace object validation can be performed once the import is completed by first exporting the contents of the target workspace:
python3 migration_pipeline.py --profile $DST_PROFILE --export-pipeline --use-checkpoint --cluster-name
And then running the validate_pipeline.sh script:
./validate_pipeline.sh $SRC_EXPORT_SESSION_ID $DST_EXPORT_SESSION_ID
Once this completes, check the console summary, as well as the logs folder (where a new folder should be generated).
Import using step-by-step tools (not recommended)
If desired, export_db.py and import_db.py can be run in a stepwise fashion. This is the legacy mode of running the tools, and in general is not recommended. If running the scripts separately, the following order of operations applies:
- Export users and groups
- Export cluster templates
- Export notebook metadata (listing of all notebooks)
- Export notebook content
- Export job templates
- Export Hive Metastore data
- Export Table ACLs
By default, artifacts are stored in the logs/ directory, and azure_logs/ for Azure artifacts.
This is configurable with the --set-export-dir flag to specify the log directory.
$ python export_db.py --help
usage: export_db.py [-h] [--users] [--workspace]
[--notebook-format {DBC,SOURCE,HTML}] [--download]
[--libs] [--clusters] [--jobs] [--metastore] [--secrets]
[--metastore-unicode] [--cluster-name CLUSTER_NAME]
[--database DATABASE] [--iam IAM] [--skip-failed]
[--mounts] [--azure] [--profile PROFILE]
[--single-user SINGLE_USER] [--export-home EXPORT_HOME]
[--export-groups EXPORT_GROUPS] [--workspace-acls]
[--workspace-top-level-only] [--silent]
[--no-ssl-verification] [--debug] [--reset-exports]
[--set-export-dir SET_EXPORT_DIR] [--pause-all-jobs]
[--unpause-all-jobs]
[--update-account-id UPDATE_ACCOUNT_ID]
[--old-account-id OLD_ACCOUNT_ID]
[--replace-old-email REPLACE_OLD_EMAIL]
[--update-new-email UPDATE_NEW_EMAIL]
[--bypass-windows-check]
Export full workspace artifacts from Databricks
optional arguments:
-h, --help show this help message and exit
--users Download all the users and groups in the workspace
--workspace Log all the notebook paths in the workspace. (metadata
only)
--notebook-format {DBC,SOURCE,HTML}
Choose the file format to download the notebooks
(default: DBC)
--download Download all notebooks for the environment
--libs Log all the libs for the environment
--clusters Log all the clusters for the environment
--jobs Log all the job configs for the environment
--metastore log all the metastore table definitions
--metastore-unicode log all the metastore table definitions including
unicode characters
--table-acls log all table ACL grant and deny statements
--cluster-name CLUSTER_NAME
Cluster name to export the metastore to a specific
cluster. Cluster will be started.
--database DATABASE Database name to export for the metastore and table
ACLs. Single database name supported
--iam IAM IAM Instance Profile to export metastore entires
--skip-failed Skip retries for any failed hive metastore exports.
--mounts Log all mount points.
--azure Run on Azure. (Default is AWS)
--profile PROFILE Profile to parse the credentials
--export-home EXPORT_HOME
User workspace name to export, typically the users
email address
--export-groups EXPORT_GROUPS
Group names to export as a set. Includes group, users,
and notebooks.
--workspace-acls Permissions for workspace objects to export
--workspace-top-level-only
Download only top level notebook directories
--silent Silent all logging of export operations.
--no-ssl-verification
Set Verify=False when making http requests.
--debug Enable debug logging
--reset-exports Clear export directory
--set-export-dir SET_EXPORT_DIR
Set the base directory to export artifacts
--pause-all-jobs Pause all scheduled jobs
--unpause-all-jobs Unpause all scheduled jobs
--update-account-id UPDATE_ACCOUNT_ID
Set the account id for instance profiles to a new
account id
--old-account-id OLD_ACCOUNT_ID
Old account ID to filter on
--replace-old-email REPLACE_OLD_EMAIL
Old email address to update from logs
--update-new-email UPDATE_NEW_EMAIL
New email address to replace the logs
$ python import_db.py --help
usage: import_db.py [-h] [--users] [--workspace] [--workspace-top-level]
[--workspace-acls] [--notebook-format {DBC,SOURCE,HTML}]
[--import-home IMPORT_HOME] [--import-groups]
[--archive-missing] [--libs] [--clusters] [--jobs]
[--metastore] [--metastore-unicode] [--get-repair-log]
[--cluster-name CLUSTER_NAME] [--skip-failed] [--azure]
[--profile PROFILE] [--single-user SINGLE_USER]
[--no-ssl-verification] [--silent] [--debug]
[--set-export-dir SET_EXPORT_DIR] [--pause-all-jobs]
[--unpause-all-jobs] [--import-pause-status]
[--delete-all-jobs] [--last-session] [--sort-views]
Import full workspace artifacts into Databricks
optional arguments:
-h, --help show this help message and exit
--users Import all the users and groups from the logfile.
--workspace Import all notebooks from export dir into the
workspace.
--workspace-top-level
Import all top level notebooks from export dir into
the workspace. Excluding Users dirs
--notebook-format {DBC,SOURCE,HTML}
Choose the file format of the notebook to import
(default: DBC)
--workspace-acls Permissions for workspace objects to import
--import-home IMPORT_HOME
User workspace name to import, typically the users
email address
--import-groups Groups to import into a new workspace. Includes group
creation and user notebooks.
--archive-missing Import all missing users into the top level /Archive/
directory.
--libs Import all the libs from the logfile into the
workspace.
--clusters Import all the cluster configs for the environment
--jobs Import all job configurations to the environment.
--metastore Import the metastore to the workspace.
--metastore-unicode Import all the metastore table definitions with
unicode characters
--table-acls Import table acls to the workspace.
--get-repair-log Report on current tables requiring repairs
--cluster-name CLUSTER_NAME
Cluster name to import the metastore to a specific
cluster. Cluster will be started.
--skip-failed Skip missing users that do not exist when importing
user notebooks
--azure Run on Azure. (Default is AWS)
--profile PROFILE Profile to parse the credentials
--no-ssl-verification
Set Verify=False when making http requests.
--silent Silent all logging of import operations.
--debug Enable debug logging
--set-export-dir SET_EXPORT_DIR
Set the base directory to import artifacts if the
export dir was a customized
--pause-all-jobs Pause all scheduled jobs
--unpause-all-jobs Unpause all scheduled jobs
--import-pause-status Import the pause status from jobs in the old workspace
--delete-all-jobs Delete all jobs
--last-session
The session to compare against. If set, the script compares current sesssion with the last session and only import updated and new notebooks.
--sort-views
Sort all views topologically based upon dependencies before importing.
e.g. if view A is created from view B, B will be imported before A. This will solve the cases when views are created from other views.
This section uses the SCIM API to export / import
user and groups.
Instance Profiles API used
to export instance profiles that are tied to user/group entitlements.
For AWS users, this section will log the instance profiles used for IAM access to resources.
To export users / groups, use the following: (The profile name DEMO will be replaced with the profile you defined for your old databricks account)
python export_db.py --profile DEMO --usersTo import these users:
python import_db.py --profile NEW_DEMO --usersIf you plan to use this tool to export multiple workspaces, you can set the --set-export-dir directory to log
artifacts into separate logging directories.
The section uses the Clusters APIs
python export_db.py --profile DEMO --clustersThis will export the following:
- Cluster templates + ACLs
- Instance pool definitions
- Cluster policies + ACLs
python import_db.py --profile NEW_DEMO --clustersIf you experience errors when you try to import the clusters, it may be that you need to modify the clusters file from the logs directory to include the new instance profile if it's not the same as the one in the old databricks account.
To make changes to a cluster name to match the new databricks account
you must edit the clusters log file after export. You do this by looking at the clusters file and identifying the old cluster instance profile which will include the old account number and the name of the instance profile.
OLD profile text from an AWS Databricks account:
arn:aws:iam::111111111111:instance-profile/profileName
The account number (111111111111) and profileName need to be found and replaced to migrate to the new account which may have a different account number and instance profile.
To modify the clusters.log file run this sed operation
sed -i 's/old-text/new-text/g' input.txtThis section uses the Workspace API
This part is a 3 part process.
- Download all notebook locations and paths
- Download all notebook contents for every path
- Download all workspace ACLs
python export_db.py --profile DEMO --workspace
python export_db.py --profile DEMO --download
python export_db.py --profile DEMO --workspace-aclsTo import into a new workspace:
python import_db.py --profile NEW_DEMO --workspace [--archive-missing]
python import_db.py --profile NEW_DEMO --workspace-acls If users have left your organization, their artifacts (notebooks / job templates) still exists. However, their user
object no longer exists. During the migration, we can keep the old users notebooks into the top level
directory /Archive/{username}@domain.com
Use the --archive-missing option to put these artifacts in the archive folder.
Single User Export/Import
The tool supports exporting single user workspaces using the following command:
# export a single users workspace
python export_db.py --profile DEMO --export-home [email protected]The corollary is the --import-home option:
python import_db.py --profile NEW_DEMO --import-home [email protected]This will include notebooks, directories, and their corresponding ACLs.
This section uses the Jobs API
Job ACLs are exported and imported with this option.
python export_db.py --profile DEMO --jobsIf we're unable to find old cluster ids that are no longer available, we'll reset the job template to use a new default cluster.
python import_db.py --profile NEW_DEMO --jobsImported jobs into the new workspace are paused by default. We do not want to have 2 jobs run simultaneously. Admins must pause their jobs with Databricks defined schedules using the following option:
python export_db.py --profile DEMO --pause-all-jobsUn-pause all jobs in the new workspace:
python import_db.py --profile NEW_DEMO --unpause-all-jobsIf you want to unpause only the jobs which were not paused in the old workspace, you can use the following option:
python import_db.py --profile NEW_DEMO --import-pause-statusThis section uses an API to remotely run Spark commands on a cluster, this API is called Execution Context
By default, this will launch a small cluster in the data/ folder to export the Hive Metastore data.
If you need a specific IAM role to export the metastore, use the --cluster-name option to connect to
a specific cluster.
By default, we will edit the cluster for every defined IAM role to loop through all failed exports in case the tool was
missing IAM permissions. To disable looping through all failed exports, you can pass in --skip-failed
# export all metastore entries and brute force loop through all instance profiles / IAM roles
python export_db.py --profile DEMO --metastore
# export all metastore entries on the default cluster without retries
python export_db.py --profile DEMO --metastore --skip-failed
# export all metastore entries on a specific cluster
python export_db.py --profile DEMO --metastore --cluster-name "Test"
# export all tables within a specific database
python export_db.py --profile DEMO --metastore --cluster-name "Test" --database "my_db"python import_db.py --profile newDEMO --metastoreTo find legacy Hive tables that need to be repaired after a successful import, run the following:
python import_db.py --profile newDEMO --get-repair-log
Once completed, it will upload a log to the destination location. Use this repair notebook to import into the destination environment to repair all tables.
The Table ACLs component includes all objects to which access is controlled using
DENY and GRANT SQL statements:
- Catalog: included if all databases are exported
- Database: included
- Table: included
- View: included (they are treated like tables with ObjectType
TABLE)
- Database: included
- Anonymous Function: included
- Any File: included
Unsupported object type:
- User Function: not included yet
This section uses the API to run notebooks on a cluster to perform the export and import. (For details, please refer to the export table ACL notebook or the import table ACL notebook)
By default, this will launch a small cluster in the data/ folder with acls suffix to export the table ACL data.
(This cluster needs to have table ACLs enabled, and it must be run with an admin user)
# export all table ACL entries
python export_db.py --profile DEMO --table-acls
# export all table ACL entries within a specific database
python export_db.py --profile DEMO --table-acls --database "my_db"For large workspaces it is not uncommon to encounter some ACLs that cause problems when
exporting: in such cases, a special log entry is made (marked with ERROR_!!!) and the export
continues. At the end error counts will be provided, and the notebooks mentioned above
contain detailed information on how to investigate any issues. Most errors are encountered
with objects that are no longer functional anyway.
This functionality exports group(s), their members, and corresponding notebooks.
This assumes an empty export directory to simplify the number of operations needed.
This does not include IAM roles as those likely change while moving across workspaces.
# reset the export directory and export a set of groups
python export_db.py --reset-export && python export_db.py --profile DEMO --export-groups 'groupA,groupB'
# import the groups that were exported
python import_db.py --profile newDEMO --import-groupsThis will export all notebooks that are not a part of the /Users/ directories to help migrate notebooks that are
outside of personal workspace directories. Usually, these will be notebooks in the '/Shared/' directory.
# reset the export directory and export the top level directories / notebooks
python export_db.py --profile DEMO --reset-export && python export_db.py --profile DEMO --workspace-top-level-only
# if ACLs are enabled, export the ACLs as well
python export_db.py --profile DEMO --workspace-acls
# import the groups that were exported
python import_db.py --profile newDEMO --workspace-top-level
# apply acls if needed
python import_db.py --profile newDEMO --workspace-aclsThis will export secret to allow migration of secrets to a new workspace.
There is a limit to the size of the secret value which will print an error if this fails.
# to export you must use a cluster
python export_db.py --profile DEMO --secrets --cluster-name "my_cluster"
# to import, you do not need a cluster
python import_db.py --profile newDEMO --secretsNote: Registered model, model version, and metric history are not supported yet. Please see mlflow-export-import for standalone MLflow migrations.
This will export and import the specified MLflow objects. Because MLflow objects depend on other object types such as workspace directories, notebooks, etc. this command should run after the other objects are successfully exported/imported.
mlflow-runs are by default only exported for the past 30 days worth of data. The user can specify other dates but should be aware of the performance impacts.
export
python3 export_db.py --profile $SRC --mlflow-experiments --use-checkpoint --num-parallel 4
python3 export_db.py --profile $SRC --mlflow-experiments-permissions --use-checkpoint --num-parallel 4
python3 export_db.py --profile $SRC --mlflow-runs --use-checkpoint --num-parallel 4 --start-date 2022-02-26
import
python3 import_db.py --profile $DST --src-profile $SRC --mlflow-experiments --use-checkpoint --num-parallel 4
python3 import_db.py --profile $DST --src-profile $SRC --mlflow-experiments-permissions --use-checkpoint --num-parallel 4
python3 import_db.py --profile $DST --src-profile $SRC --mlflow-runs --use-checkpoint --num-parallel 4
Note: To disable ssl verification pass the flag --no-ssl-verification.
If still getting SSL Error add the following to your current bash shell:
export REQUESTS_CA_BUNDLE=""
export CURL_CA_BUNDLE=""
- Instance profiles (AWS only): Group access to instance profiles will take precedence. If a user is added to the role directly, and has access via a group, only the group access will be granted during a migration.
- Clusters: Cluster creator will be seen as the single admin user who migrated all the clusters. (Relevant for billing
purposes)
- Cluster creator tags cannot be updated. Added a custom tag named
OriginalCreatorwith the original cluster creator for DBU tracking.
- Cluster creator tags cannot be updated. Added a custom tag named
- Jobs: Job owners will be seen as the single admin user who migrate the job configurations. (Relevant for billing
purposes)
- Jobs with existing clusters that no longer exist will be reset to the default cluster type
- Jobs with older legacy instances will fail with unsupported DBR or instance types. See release notes for the latest supported releases.