fix(postgres-source): stop pgjdbc confirming the slot, and rebound the confirmation test - #5976
Merged
jarohen merged 2 commits intoSep 1, 2026
Merged
Conversation
Copy link
Copy Markdown
Member
Sorry, something went wrong.
Closed
4 tasks
…, not the block (resolves xtdb#5974) Why the block token was the wrong bound, and the measurement behind the new one, are on xtdb#5974. - **The regression the test exists for still fails it.** Confirmation tracking apply, which is what xtdb#5867 fixed, reaches the third row's commit — comfortably past the sample. - **The closing assertion moves from `> blockLsn` to `>= persisted`.** It read as "the slot advances once the block is durable", but the quiet advance had already made it true before the second block existed, so it asserted nothing. Comparing against the new block's own token is the claim that was meant. Refs xtdb#5974 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
…esolves xtdb#5975) The mechanism, and the loss it opens at stream open, are on xtdb#5975. - **The spec gains a line on `SoleConfirmer` rather than a new obligation.** "At most one party confirms a given slot" already covers this; what it did not say is that a driver advancing the flush position on its own schedule counts as a party. A second invariant naming the driver case would oblige the same thing twice, and a reader checking coverage would count it twice. - **No regression test comes with it.** Reproducing the loss needs a keepalive to land between stream open and the first data message, on a source resuming with an un-blocked range behind it; that timing is not constructible from the client. What is testable is the other direction, and `slot advances before the first block is cut` still passes with automatic flush off — so the quiet advance is `confirm()`'s alone, and no guarantee was resting on the driver. Refs xtdb#5975 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
jarohen
force-pushed
the
pgsrc-confirmation-5974
branch
from
174df88 to
2016b35
Compare
September 1, 2026 15:44
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
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.
Resolves #5974 and #5975.
slot is confirmed only as far as the last durable block failed twice on main in one day with a green run between them, leaving the Integration Test job intermittently red for every branch rebasing onto it. #5975 was found while investigating that, in the same keepalive path.