Repository navigation
[trino] Add native read-only Trino connector #4197
Description
Search before asking
- I searched in the issues and found nothing similar.
Description
Motivation
Apache Fluss currently has no native Trino connector for querying Fluss-managed data directly.
Previous work in #1810 and FIP-18 explored a broader Trino integration, including Lakehouse Union Read and multi-version support.
This issue tracks a narrower first step: a native, read-only Trino connector for Fluss.
Scope
The initial connector focuses on:
- metadata discovery and type mapping;
- bounded reads of non-Lakehouse Log Tables;
- current-state reads of non-Lakehouse Primary Key Tables;
- projection and predicate-based scan optimizations;
- partition/bucket pruning and primary-key lookup where applicable.
Lakehouse Union Read, Lakehouse format readers, writes/DDL, and Trino core changes are intentionally out of scope.
Design
Detailed design and discussion:
- FIP-XXX: Native Read-Only Trino Connector for Apache Fluss
- Design doc: https://docs.google.com/document/d/1mY-PN8KY1cJSD9YzISuL14A6rEIH1Htj9l99XrLmPk0
- Discussion thread: https://lists.apache.org/thread/275nbvz6w2288ky9mq70962nscvxm54o
Prior Work
PR track
- Initial trino connector [trino] Add initial connector build and plugin skeleton #4207
- Native Fluss client bounded scan [Client] Support bounded log scanning #4545
- Initial, limited support for partition pushdown. [trino] Add predicate constraint handling and equality-based partition pruning #4689
Issue Track
- Support bounded Log Table scanning [client] Support bounded Log Table scanning #4544
- Log fetch may get stuck when EARLIEST resolves to a lake-only offset [Bug] Log fetch may get stuck when EARLIEST resolves to a lake-only offset #4676
- Make log_start_offset authoritative in FetchLog [RPC] Make log_start_offset authoritative in FetchLog v1 #4677
Willingness to contribute
- I'm willing to submit a PR!
Activity
polyzos commentedon Sep 2, 2026
@yangshangqing95 thanks a lot for driving this.. i have seen some requests for a fluss-trino connector, so this is very much well needed 🙌
yangshangqing95 commentedon Sep 2, 2026
Hi @polyzos Thanks! Glad to hear there’s interest in this.
Based on the current design, I’m planning to start with the native read-only Fluss path first, while keeping the design open to feedback as the implementation progresses.
At the same time, I'll keep the design open and continue refining it based on feedback from the community as the implementation progresses, especially around module/build structure and the longer-term Union Read direction.
I'll try to keep the work incremental and split it into small, reviewable PRs.
polyzos commentedon Sep 4, 2026
@yangshangqing95 did you manage to get access to Confluence?
yangshangqing95 commentedon Sep 4, 2026
HI @polyzos I applied for days ago in https://selfserve.apache.org/confluence-account.html, but not received any response yet.
polyzos commentedon Sep 4, 2026
@yangshangqing95 can you try submitting again? I tried approving it when the request came and now again, but for some reason the ASF page throws an error 🫠
yangshangqing95 commentedon Sep 4, 2026
@polyzos Thanks for your supporting!
I just submitted with another email address, as the old one shows
polyzos commentedon Sep 4, 2026
I keep hitting this error @wuchong @fresh-borzoni @leekeiabstraction any chance it works for you? or we need to open a support ticket?
yangshangqing95 commentedon Sep 4, 2026
Thanks a lot for your support and for looking into this so quickly. @polyzos Much appreciated!
In the meantime, while we wait for the Confluence issue to be resolved, maybe we can continue making progress in parallel. I’ve already submitted a PR for the basic connector skeleton, and we could also start the design discussion based on the Google Doc?
Once Confluence is available again, I can move the design, along with the conclusions from the discussion, into the formal FIP page.
Hopefully this lets us keep the work moving without making the Confluence issue a blocker. cc @wuchong @fresh-borzoni @leekeiabstraction
I’d be very happy to hear everyone’s thoughts on the proposal.
Also, if anyone has some time, I’d really appreciate a review of the PR above. It establishes the basic foundation for the follow-up work. I’ve intentionally kept the change small and isolated, so it should not affect the existing modules or introduce behavior changes elsewhere in Fluss.
yangshangqing95 commentedon Sep 9, 2026
Hi @polyzos, following up on this since the Confluence access issue is still unresolved.
The design in Google doc and initial connector skeleton PR is also ready for review, so any feedback there would be very helpful for moving the work forward.
polyzos commentedon Sep 10, 2026
@yangshangqing95 I will add some comments on the mailing list for visibility.
Regarding Confluence access, it seems like this is an ASF infra issue, so we are also waiting for it to be addressed