DataHub Quickstart Guide
To deploy a new instance of DataHub, perform the following steps.
- Install Docker and Docker Compose v2 for your platform.
Make sure to allocate enough hardware resources for Docker engine. Tested & confirmed config: 2 CPUs, 8GB RAM, 2GB Swap area, and 10GB disk space.
Launch the Docker Engine from command line or the desktop app.
Install the DataHub CLI
a. Ensure you have Python 3.7+ installed & configured. (Check using
b. Run the following commands in your terminal
python3 -m pip install --upgrade pip wheel setuptools
python3 -m pip uninstall datahub acryl-datahub || true # sanity check - ok if it fails
python3 -m pip install --upgrade acryl-datahub
If you see "command not found", try running cli commands with the prefix 'python3 -m' instead like
python3 -m datahub version
Note that DataHub CLI does not support Python 2.x.
To deploy a DataHub instance locally, run the following CLI command from your terminal
datahub docker quickstart
This will deploy a DataHub instance using docker-compose. If you are curious, the
docker-compose.yamlfile is downloaded to your home directory under the
If things go well, you should see messages like the ones below:
Fetching docker-compose file https://raw.githubusercontent.com/datahub-project/datahub/master/docker/quickstart/docker-compose-without-neo4j-m1.quickstart.yml from GitHub
Pulling docker images...
Finished pulling docker images!
[+] Running 11/11
⠿ Container zookeeper Running 0.0s
⠿ Container elasticsearch Running 0.0s
⠿ Container broker Running 0.0s
⠿ Container schema-registry Running 0.0s
⠿ Container elasticsearch-setup Started 0.7s
⠿ Container kafka-setup Started 0.7s
⠿ Container mysql Running 0.0s
⠿ Container datahub-gms Running 0.0s
⠿ Container mysql-setup Started 0.7s
⠿ Container datahub-datahub-actions-1 Running 0.0s
⠿ Container datahub-frontend-react Running 0.0s
✔ DataHub is now running
Ingest some demo data using `datahub docker ingest-sample-data`,
or head to http://localhost:9002 (username: datahub, password: datahub) to play around with the frontend.
Need support? Get in touch on Slack: https://slack.datahubproject.io/
Upon completion of this step, you should be able to navigate to the DataHub UI at http://localhost:9002 in your browser. You can sign in using
datahubas both the username and password.
On Mac computers with Apple Silicon (M1, M2 etc.), you might see an error like
no matching manifest for linux/arm64/v8 in the manifest list entries, this typically means that the datahub cli was not able to detect that you are running it on Apple Silicon. To resolve this issue, override the default architecture detection by issuing
datahub docker quickstart --arch m1
To ingest the sample metadata, run the following CLI command from your terminal
datahub docker ingest-sample-data
If you've enabled Metadata Service Authentication, you'll need to provide a Personal Access Token
--token <token> parameter in the command.
That's it! Now feel free to play around with DataHub!
Command not found: datahub
If running the datahub cli produces "command not found" errors inside your terminal, your system may be defaulting to an
older version of Python. Try prefixing your
datahub commands with
python3 -m datahub docker quickstart
Another possibility is that your system PATH does not include pip's
$HOME/.local/bin directory. On linux, you can add this to your
if [ -d "$HOME/.local/bin" ] ; then
By default the quickstart deploy will require the following ports to be free on your local machine:
3306 for MySQL
9200 for Elasticsearch
9092 for the Kafka broker
8081 for Schema Registry
2181 for ZooKeeper
9002 for the DataHub Web Application (datahub-frontend)
8080 for the DataHub Metadata Service (datahub-gms)
In case the default ports conflict with software you are already running on your machine, you can override these ports by passing additional flags to the
datahub docker quickstartcommand. e.g. To override the MySQL port with 53306 (instead of the default 3306), you can say:
datahub docker quickstart --mysql-port 53306. Use
datahub docker quickstart --helpto see all the supported options. For the metadata service container (datahub-gms), you need to use an environment variable,
DATAHUB_MAPPED_GMS_PORT. So for instance to use the port 58080, you would say
DATAHUB_MAPPED_GMS_PORT=58080 datahub docker quickstart
no matching manifest for linux/arm64/v8 in the manifest list entries
Miscellaneous Docker issues
docker system prune
To add users to your deployment to share with your team check out our Adding Users to DataHub
To enable backend Authentication, check out [authentication in DataHub's backend](/docs/authentication/introducing-metadata-service-authentication#Configuring Metadata Service Authentication).
Move to Production
We recommend deploying DataHub to production using Kubernetes. We provide helpful Helm Charts to help you quickly get up and running. Check out Deploying DataHub to Kubernetes for a step-by-step walkthrough.
Other Common Operations
To stop DataHub's quickstart, you can issue the following command.
datahub docker quickstart --stop
Resetting DataHub (a.k.a factory reset)
To cleanse DataHub of all of its state (e.g. before ingesting your own), you can use the CLI
datahub docker nuke
Backing up your DataHub Quickstart (experimental)
The quickstart image is not recommended for use as a production instance. See Moving to production for recommendations on setting up your production cluster. However, in case you want to take a backup of your current quickstart state (e.g. you have a demo to your company coming up and you want to create a copy of the quickstart data so you can restore it at a future date), you can supply the
--backup flag to quickstart.
datahub docker quickstart --backup
will take a backup of your MySQL image and write it by default to your
~/.datahub/quickstart/ directory as the file
backup.sql. You can customize this by passing a
datahub docker quickstart --backup --backup-file /home/my_user/datahub_backups/quickstart_backup_2002_22_01.sql
Note that the Quickstart backup does not include any timeseries data (dataset statistics, profiles, etc.), so you will lose that information if you delete all your indexes and restore from this backup.
Restoring your DataHub Quickstart (experimental)
As you might imagine, these backups are restore-able. The following section describes a few different options you have to restore your backup.
Restoring a backup (primary + index) [most common]
To restore a previous backup, run the following command:
datahub docker quickstart --restore
This command will pick up the
backup.sql file located under
~/.datahub/quickstart and restore your primary database as well as the elasticsearch indexes with it.
To supply a specific backup file, use the
datahub docker quickstart --restore --restore-file /home/my_user/datahub_backups/quickstart_backup_2002_22_01.sql
Restoring only the index [to deal with index out of sync / corruption issues]
Another situation that can come up is the index can get corrupt, or be missing some update. In order to re-bootstrap the index from the primary store, you can run this command to sync the index with the primary store.
datahub docker quickstart --restore-indices
Restoring a backup (primary but NO index) [rarely used]
Sometimes, you might want to just restore the state of your primary database (MySQL), but not re-index the data. To do this, you have to explicitly disable the restore-indices capability.
datahub docker quickstart --restore --no-restore-indices
Upgrading your local DataHub
If you have been testing DataHub locally, a new version of DataHub got released and you want to try the new version then you can just issue the quickstart command again. It will pull down newer images and restart your instance without losing any data.
datahub docker quickstart
If you would like to customize the DataHub installation further, please download the docker-compose.yaml used by the cli tool, modify it as necessary and deploy DataHub by passing the downloaded docker-compose file:
datahub docker quickstart --quickstart-compose-file <path to compose file>