Recover service workers from transient Server outages

CHANGELOG.md

@@ -1,5 +1,14 @@
 # Changelog
 
+## 3.2.1
+
+- Add `Worker::recover_transient_outages(true)` for long-running service workers.
+  Retryable poll and worker-heartbeat failures keep retrying with capped backoff
+  and the original poll request identity. Shutdown interrupts retry waits.
+- Preserve bounded retries by default and for `run_once`. Disabling ordinary
+  retries still takes precedence. Authentication, protocol, codec, handler and
+  task settlement failures remain errors.
+
 ## 3.2.0
 
 - Add typed worker-session options and shared create, renew and close handles.

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "durable-workflow"
-version = "3.2.0"
+version = "3.2.1"
 edition = "2021"
 description = "Rust client and worker SDK for Durable Workflow Cloud and self-hosted Server"
 license = "MIT"
@@ -14,7 +14,7 @@ categories = ["api-bindings", "asynchronous"]
 include = ["/src/**", "/schema/**", "/examples/**", "/Cargo.toml", "/README.md", "/CHANGELOG.md", "/LICENSE"]
 
 [package.metadata.durable-workflow]
-product-train = "3.2.0"
+product-train = "3.2.1"
 compatibility-authority = "protocol-manifests"
 supported-server-versions = "2.0.0"
 qualified-server-version = "2.5.1"

README.md

@@ -151,6 +151,20 @@ The runnable session example uses a real process-local cache and prints its reso
 generation. Use Server 2.5.1 / Native 2.4.1 for session history and original TTL
 preservation. Session support starts with SDK 3.2.0.
 
+### Recovering from Server outages
+
+For a long-running service worker, enable `.recover_transient_outages(true)`
+on `Worker`. Retryable poll and worker-heartbeat failures then keep retrying
+with capped exponential backoff. A retried poll keeps its original request ID.
+`run_until` interrupts retry waits when shutdown arrives and settles any poll
+response already in flight.
+
+The default preserves bounded retries. `run_once` remains bounded in either
+mode, and `WorkerRetryPolicy.max_retries = 0` disables ordinary retries.
+Authentication, protocol, codec, handler and task settlement failures still
+return an error. Registration and deregistration are outside this recovery
+option. Keep a process supervisor for startup failures and process crashes.
+
 ### Runnable examples
 
 | Example | Demonstrates |