Durable emitter - error handling

[tarcisiozf]

Requires

Supports

[github-actions]

👋 tarcisiozf, thanks for creating this pull request!

To help reviewers, please consider creating future PRs as drafts first. This allows you to self-review and make any final changes before notifying the team.

Once you're ready, you can mark it as "Ready for review" to request feedback. Thanks!

[tarcisiozf]

waiting for #2031 to be merged

[github-actions]

⚠️ API Diff Results - github.com/smartcontractkit/chainlink-common

⚠️ Breaking Changes (1)

pkg/beholder.DurableEventStore (1)

• MarkFailedBatch — ➕ Added

---

📄 View full apidiff report

[Copilot]

Pull request overview

This PR attempts to add partial PublishBatch error handling for the durable emitter by enabling chipingress partial-success responses and recording failed event batches in the durable store.

Changes:

• Adds MarkFailedBatch to the durable event store interface and metrics wrapper.
• Updates typed batch publishing to request partial success and process per-event errors.
• Points the root module at the local pkg/chipingress module.

Reviewed changes

Copilot reviewed 4 out of 5 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
pkg/beholder/durable_event_store.go	Adds failure-recording API support and metrics instrumentation.
pkg/beholder/durable_emitter.go	Enables typed PublishBatch partial-success handling and records partial failures.
pkg/beholder/durable_emitter_test.go	Adds a no-op MarkFailedBatch implementation to the in-memory test store.
go.mod	Adds a local replace for the chipingress module.
go.sum	Removes chipingress module checksums now hidden by the local replace.

Comments suppressed due to low confidence (2)

pkg/beholder/durable_emitter.go:804

• After recording per-event failures, this path still returns nil, so flushBatch treats the whole PublishBatch as successful and calls MarkDeliveredBatch for every ID in the batch. Any events present in response.Results with errors will be marked delivered/deleted instead of remaining pending for retransmit, which defeats the failure tracking added here.

	if len(groupByError) > 0 {
		if err := d.reportPartialFailures(ctx, groupByError); err != nil {
			return fmt.Errorf("publish errors reporting failed: %w", err)
		}
	}

	return nil

pkg/beholder/durable_emitter.go:795

• If the server returns an errored result for an event ID that is not in the batch, continuing here lets flushBatch mark the entire batch delivered even though the response could not be reconciled to stored rows. Treat an unknown errored event_id as a publish failure for the batch so the original rows remain pending instead of risking data loss.

		id, exists := mapEventIDtoTableId[result.EventId]
		if !exists {
			d.log.Warnw("PublishBatch returned invalid event", "event_id", result.EventId)
			continue
		}

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.