Frequently asked questions
Each question links to the page that answers it.
Evaluate
- What is a metalake, and what does it keep separate?
- What do catalog, schema, and metalake mean in Gravitino?
- How does a request flow from an engine through Gravitino to the data?
- Which sources can I connect?
- Does the Kafka catalog work with Redpanda or Amazon MSK?
- Does Gravitino store or copy my data?
- How large can a deployment grow?
- Which third-party software ships with Datastrato Enterprise?
Install and upgrade
- What do I need before I install?
- Which ports does each service use?
- How do I install on EKS, AKS, GKE, or OpenShift?
- Can I install without internet access?
- How do I get and apply a license key?
- What happens when the license expires?
- How do I keep passwords out of my values file?
- Which database holds Gravitino's own metadata?
- How do I make a deployment highly available?
- Can I run more than one server?
- How do I upgrade to a new release?
- How do I roll back an upgrade?
- What should I back up?
- What does an uninstall delete?
- What should be in place before going to production?
Sign in and security
- Which sign-in methods are supported?
- How do I connect my identity provider?
- How do users and groups get from Okta or Entra ID into Gravitino?
- How do I create local users and reset a password?
- How do I turn on HTTPS, and do I need a service mesh?
- Is traffic inside the cluster encrypted?
- How do I harden a deployment?
- Why was I signed out of the UI?
Access control
- How do I give a user access to a table?
- Which privileges does each operation need?
- How do I see who can access what?
- How do allow and deny grants combine?
- Can Gravitino push its grants down to Ranger?
Connect engines
- Should an engine use a Gravitino connector or the Iceberg REST catalog?
- Where do I find the endpoints and connector downloads for my deployment?
- How do I install the Trino connector?
- How do I use the Spark connector?
- How do I add connectors to Trino or Spark running on Kubernetes?
- How do I get the Daft connector?
- How do engines get credentials for object storage?
Iceberg REST catalog
- How do I point Spark, Trino, Flink, or PyIceberg at the Iceberg REST catalog?
- Can I have more than one Iceberg catalog?
- How does a client choose a catalog?
- Can Gravitino's metadata and the Iceberg catalog's metadata use the same database?
- Which Iceberg REST operations are supported?
- How do I use catalogs on more than one cloud?
- Does the Iceberg REST catalog support server-side scan planning?
- Does the Iceberg REST catalog cache table metadata?
- How do I federate remote Iceberg REST catalogs into my Iceberg REST catalog?
- Does a federated catalog have to trust the same identity provider?
- What happens to the files when I drop an Iceberg table?
- How do I compact Iceberg tables?
Lance
Governance
- How do I tag tables and columns?
- What is a policy, and what can it apply to?
- How do I collect lineage from Spark?
- What are statistics, and how do they differ from metrics?
- How do AI agents work with Gravitino?
Operate
- How do I check that the server is healthy?
- How do I connect Prometheus, and which metrics are there?
- Where is the server log?
- Who did what, and when?
- How do I restart or stop a component?
- How do I send catalog changes to another system?