DanielLeens commented on issue #9866:
URL: https://github.com/apache/seatunnel/issues/9866#issuecomment-5137561182
I rechecked this MongoDB CDC sharded-split failure against the current code
and the new in-flight fix.
The original stack trace is consistent with a real bug in the sharded
metadata path. `ShardedSplitStrategy` expects `config.collections.key` to be
available when building the shard-key row type and chunk boundaries, but the
current metadata query does not project that `key` field. As a result, a valid
sharded collection can still reach `collectionMetadata.getDocument("key")` with
the field missing and fail with `Document does not contain key 'key'`.
There is now a focused draft fix PR: #11599. It changes the MongoDB metadata
projection to include the shard key field and adds a unit test for that
projection. The PR is still draft and its current Build check is failing, so
this issue should stay open until the PR is reviewed, CI is green, and the fix
lands in `dev`.
After #11599 is merged, the best verification is to rerun the same MongoDB
4.4 sharded-cluster job and confirm that split generation no longer fails
before snapshot assignment. If it still reproduces on a build containing that
PR, please update this issue with the sanitized `config.collections` metadata
shape for the collection and the new stack trace.
--
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]
