fix(connectors): propagate sink plugin consume failure status

[minleejae]

Binds the iggy_sink_consume FFI status and returns RuntimeError::SinkConsumeFailed { plugin_id, status } on non-zero, matching the existing iggy_sink_open check in the same file. The caller's existing Err path handles logging and skips the processed counter, so a failed batch is no longer counted as processed.

Includes a regression test with stub extern "C" callbacks for both paths — the failure-path test fails on master with expected SinkConsumeFailed, got success and passes with this change.

Closes #3950

[github-actions]

Thanks for the PR. It is labeled S-waiting-on-review and queued for review.

Slash commands (own line, regular comment) move it around the queue:

• /ready - back to S-waiting-on-review after addressing feedback
• /author - flip to S-waiting-on-author while you finish changes
• /request-review @user-or-team - request a reviewer

See CONTRIBUTING.md for details.

[kriti-sc]

while I agree that the plugin should exit when plugin returns non-zero status, simply exitting the plugin does not solve the problem highlighted in the issue.

I think we should fix the comment and update the issue description to reflect this.

Also see discussion: #2927

[minleejae]

You're right — reworded the comment and updated the issue description: this change is fail-fast (surfaces the failure, stops counting failed batches as processed), not redelivery, since offsets are already committed at poll time. Happy to adjust this PR to whatever design lands in #2927.

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs.

If you need a review, please ensure CI is green and the PR is rebased on the latest master. Don't hesitate to ping the maintainers - either @core on Discord or by mentioning them directly here on the PR.

Thank you for your contribution!

[hubcio]

closing in favour of #3954, which does the same status bind at sink.rs:740 plus the offset-commit half you scoped out of here - so a rejected batch isn't just reported, it's redelivered.

nothing wrong with this patch; #3950's repro is what got the runtime fix prioritized. flagged on #3954 that your process_messages unit tests should carry over - it only has integration coverage of that path.

#3950 stays open until #3954 merges. thanks for the repro. @kriti-sc