davidzollo opened a new issue, #11051:
URL: https://github.com/apache/seatunnel/issues/11051

   ## Background
   SeaTunnel's `MongoDB-CDC` source currently lacks a reliable first-class 
startup mode for "start from now and consume incremental changes only".
   
   There is an existing user report showing that attempts to run in an 
incremental-only style still fall back to snapshot behavior, which makes repeat 
job submission unsafe for users who only want new changes.
   
   This issue is intended to track the engineering work needed to add a clear, 
contributor-friendly implementation target for that startup mode.
   
   ## Scope
   Add a `MongoDB-CDC` startup mode that:
   - skips the initial snapshot
   - starts from the latest available change-stream position
   - continues consuming incremental changes only
   
   This issue is about **latest-offset / incremental-only startup semantics**. 
It does **not** cover dynamic newly-added collection capture or new metadata 
fields.
   
   ## Suggested implementation approach
   ### 1. Make the startup mode explicit in connector-owned options
   - Expose the mode through `MongodbIncrementalSourceOptions` rather than 
relying on users to combine loosely related settings.
   - Keep configuration validation aligned with actual runtime support.
   - Update stale option descriptions so docs, validation, and behavior match.
   
   ### 2. Route startup behavior through one clear source path
   - When latest-offset mode is selected, the connector must skip snapshot/copy 
logic entirely.
   - The source should initialize from the latest change-stream position that 
matches the intended semantics of "new changes only".
   - Snapshot-related options should either be rejected or documented as 
ignored in this mode.
   
   ### 3. Preserve checkpoint / restore correctness
   - Startup-mode selection must flow through the same checkpoint / restore 
path as the existing MongoDB source.
   - Restart should resume from checkpoint state rather than accidentally 
re-entering a snapshot path.
   
   ### 4. Tests and validation
   Suggested test layers:
   - option parsing / validation tests
   - startup-path tests proving snapshot is skipped
   - restart / restore tests proving the mode remains incremental-only after 
recovery
   - doc examples in both English and Chinese
   
   ## Related history
   - User report: #9511
   
   ## Suggested acceptance criteria
   - `MongoDB-CDC` exposes an explicit latest-offset / incremental-only startup 
mode.
   - Snapshot is skipped when this mode is selected.
   - Restart / restore does not silently fall back to snapshot behavior.
   - Option docs and runtime validation are consistent.
   - English and Chinese docs are updated.
   - Add focused tests for parsing, startup-path selection, and recovery 
behavior.
   
   ## Non-goals
   - Dynamic newly-added collection discovery.
   - New metadata keys.
   - General CDC transform enhancements.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to