Skip to content

Free, Open Source, Self-Hosted WhatsApp API Gateway

docs: describe the redrive binding filter and the catalog entry checks

Yudhi Armyndharis1f111cd2a66a
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7bea6fd56..bc0c18ca5 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -309,6 +309,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- Refuse a plugin catalog entry whose id, name, version or text fields are not text, and accept null in its optional fields.
- Log once each `SSRF_ALLOWED_HOSTS` entry that carries a scheme, port, path or wildcard, which never matches a host.
- Run every `message:sending` handler, a moderation veto included, for a send that a hook handler schedules after its own call has ended; it was skipped as re-entrant.
+- Redrive only the dead letters written under an ingress instance's current session binding.
+- Drop a null optional field from plugin catalog entries instead of answering it.
### Dependencies
diff --git a/docs/06-api-specification.md b/docs/06-api-specification.md
index 795b77532..7ceb7e534 100644
--- a/docs/06-api-specification.md
+++ b/docs/06-api-specification.md
@@ -6131,7 +6131,7 @@ The catalog is fetched from `PLUGIN_CATALOG_URL`, which defaults to the OpenWA-p
The dashboard installs and updates an entry by passing its `download` URL to `POST /api/plugins/install-url` or `POST /api/plugins/:id/update`, so the pin rule of those routes applies: under `NODE_ENV=production` (unless `PLUGIN_INSTALL_REQUIRE_PIN=false`) an entry whose `download` has no `#sha256=` pin cannot be installed. Every entry in the default catalog carries one.
-**Errors:** `400` catalog fetch failed / not a JSON array · `401` · `403`
+**Errors:** `400` catalog fetch failed / not a JSON array / an entry without a text `id`, `name` or `version`, or with a non-text optional field · `401` · `403`
---
@@ -6452,7 +6452,8 @@ Re-dispatch one bounded batch of dead-lettered inbound deliveries for an integra
**doc 25, Integration Fabric** for the DLQ and redrive design). Rows are drained fewest-attempts-first,
then oldest-first, up to `batchSize` (100) — so a row that keeps failing moves behind never-retried
rows instead of livelocking the window, while staying redrivable. `remaining` reports the DLQ depth
-still outstanding after this batch.
+still outstanding after this batch. Only rows written under the instance's current session binding are
+redriven and counted; rows of an earlier binding stay in the dead-letter table.
**Auth:** API key (ADMIN) · **Scope:** session-scoped (a key restricted to `allowedSessions` may only
redrive an instance whose current `sessionScope` is inside that allowlist; out of scope and missing
diff --git a/docs/25-integration-fabric.md b/docs/25-integration-fabric.md
index 7f39993b2..a0859f58d 100644
--- a/docs/25-integration-fabric.md
+++ b/docs/25-integration-fabric.md
@@ -264,8 +264,9 @@ few seconds.
A queued, retried, reconciled or redriven delivery runs with what was decided when it was accepted: the
verification verdict (a delivery stored before the verdict was recorded reads the route as it stands) and
the session its instance was bound to. If the instance has been disabled or deleted, or bound to another
-session, by the time the delivery runs, the attempt fails and follows the retry and dead-letter path; a
-redrive of such a row fails the same way.
+session, by the time the delivery runs, the attempt fails and follows the retry and dead-letter path. A
+redrive covers only the rows written under the instance's current session binding; rows of an earlier
+binding stay in the dead-letter table.
A job waiting on that lock still holds one of the `INGRESS_WORKER_CONCURRENCY` worker slots (default
10). A burst on one lane larger than that fills every slot with same-lane waiters, and events for other