[cloudflare] add docs for the durable objects Queue

Author: conico974

pages/cloudflare/caching.mdx

@@ -5,9 +5,9 @@ import { Callout } from "nextra/components";
 
 `@opennextjs/cloudflare` supports [caching](https://nextjs.org/docs/app/building-your-application/data-fetching/fetching-caching-and-revalidating#caching-data).
 
-By default, all `fetch()` subrequests made in your Next.js app are cached. Refer to the [Next.js documentation](https://nextjs.org/docs/app/building-your-application/caching#opting-out-1) for information about how to disable caching for an individual subrequest, or for an entire route.
+In Next.js 14, all `fetch()` requests made in your app are cached by default. However, in Next.js 15, `fetch()` requests are not cached by default. Refer to the [Next.js documentation](https://nextjs.org/docs/app/building-your-application/caching#opting-out-1) for information about how to enable or disable caching for an individual request, or for an entire route.
 
-[The cache persists across deployments](https://nextjs.org/docs/app/building-your-application/caching#data-cache). You are responsible for revalidating/purging this cache.
+[The cache is unique per build (prefixed by BUILD_ID) so caches are not reused between deployments](https://nextjs.org/docs/app/building-your-application/caching#data-cache). You are responsible for revalidating/purging the cache for a specific build or if you keep the same build ID.
 
 Next.js primes the cache at build time. The build time values are serverd by the [Workers Assets](https://developers.cloudflare.com/workers/static-assets/).
 
@@ -57,28 +57,57 @@ The binding name used in your app's worker is `NEXT_CACHE_WORKERS_KV`.
 
 In your project's OpenNext config, enable the KV cache and set up a queue.
 
-The memory queue will send revalidation requests to a page when needed, and offers support for de-duplicating requests on a per-isolate basis. There might still be duplicate requests under high traffic or across regions.
-
-<Callout type="warning">
-  The memory queue provided by `@opennextjs/cloudflare` is not fully suitable for production deployments, you
-  can use it at your own risk!
-</Callout>
+The Durable Object Queue will send revalidation requests to a page when needed, and offers support for de-duplicating requests.
+By default there will be a maximum of 10 instance of the Durables Object Queue and they can each process up to 5 requests in parallel.(For up to 50 ISR revalidations in parallel)
 
 ```ts
 // open-next.config.ts
 import { defineCloudflareConfig } from "@opennextjs/cloudflare";
 import kvIncrementalCache from "@opennextjs/cloudflare/kv-cache";
-import memoryQueue from "@opennextjs/cloudflare/memory-queue";
+import doQueue from "@opennextjs/cloudflare/durable-queue";
 
 export default defineCloudflareConfig({
   incrementalCache: kvIncrementalCache,
-  queue: memoryQueue,
+  queue: doQueue,
 });
 ```
 
+You will also need to add some binding to your `wrangler.jsonc` file.
+
+```jsonc
+"durable_objects": {
+    "bindings": [
+      {
+        "name": "NEXT_CACHE_REVALIDATION_DURABLE_OBJECT",
+        "class_name": "DurableObjectQueueHandler"
+      }
+    ]
+  },
+  "migrations": [
+    {
+      "tag": "v1",
+      "new_sqlite_classes": ["DurableObjectQueueHandler"]
+    }
+  ],
+```
+
+You can customize the behaviors of the queue with environment variables:
+- The max number of durable workers that can be created (MAX_REVALIDATE_CONCURRENCY)
+- The max number of revalidations that can be processed by an instance of durable worker at the same time (MAX_REVALIDATION_BY_DURABLE_OBJECT)
+- The max time in milliseconds that a revalidation can take before being considered as failed (REVALIDATION_TIMEOUT_MS)
+- The amount of time after which a revalidation will be attempted again if it failed. If it fails again it will exponentially back off until it reaches the max retry interval (REVALIDATION_RETRY_INTERVAL_MS)
+- The maximum number of attempts that can be made to revalidate a path (MAX_REVALIDATION_ATTEMPTS)
+
+
 <Callout>
-  The `direct` mode for the queue is intended for debugging purposes and is not recommended for use in
+  There is 2 additional modes that you can use for the queue `direct` and the memory queue
+
+- The memory queue will dedupe request but only on a per isolate basis. It is not fully suitable for production deployments, you
+  can use it at your own risk!
+
+- The `direct` mode for the queue is intended for debugging purposes and is not recommended for use in
   production. We are actively working on a solution that will be suitable for production.
+
 </Callout>
 
 #### On-Demand Revalidation