Add a Databricks toolset for Unity Gateway MCP Services - #74198
Open
firasbouzazi wants to merge 5 commits into
Open
firasbouzazi wants to merge 5 commits into
firasbouzazi wants to merge 5 commits into
Conversation
Dag authors want their agents to use governed Databricks tools such as Unity Catalog functions, Genie and AI Search through a Unity AI Gateway MCP Service, without putting bearer tokens or gateway URLs in Dag code. The generic MCP toolset needs both in a separate connection and sends a fixed token, which expires during long agent runs with OAuth identities. closes: apache#74193
|
Congratulations on your first Pull Request and welcome to the Apache Airflow community! If you have any issues or are unsure about any anything please check our Contributors' Guide
|
Each gateway request fetched its token through AirflowToolset.run_blocking, whose lock is shared by every toolset in the process. While another toolset ran a long blocking call, such as a SQL query, every request to the MCP Service waited for it to finish.
Databricks requires USE CATALOG and USE SCHEMA on the parent catalog and schema as well as EXECUTE on the service, so an identity granted only EXECUTE is denied. Users following the guide or the access-denied error would grant too little.
Databricks also requires the calling identity to be assigned to the workspace the request goes to, which the guide did not say.
The MCP client carries on after some gateway error responses, such as one to a notification, but the toolset kept that status and later reported an unrelated failure as it. A tool call with bad arguments then ended the agent run as a gateway error instead of letting the model retry. Also use Databricks' name for the product, Unity Gateway, and correct the guide: Databricks provides MCP Services for workspace tools and SaaS applications, and only external MCP servers can be registered as one.
firasbouzazi
changed the title
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Dag authors want their agents to use governed Databricks tools such as Unity Catalog functions, Genie and AI Search through a Unity AI Gateway MCP Service, without putting bearer tokens or gateway URLs in Dag code. The generic MCP toolset needs both in a separate connection and sends a fixed token, which expires during long agent runs with OAuth identities.
closes: #74193
Was generative AI tooling used to co-author this PR?
{pr_number}.significant.rst, in airflow-core/newsfragments. You can add this file in a follow-up commit after the PR is created so you know the PR number.