NLB-7870: Provide guidance on upgrading to NGINX Plus 37.0.4

content/nginxaas-azure/changelog/changelog.md

@@ -10,10 +10,17 @@ url: /nginxaas-azure/changelog/
 
 Learn about the latest updates, new features, and resolved bugs in F5 NGINXaaS for Azure.
 
-To see a list of currently active issues, visit the [Known issues]({{< ref "/nginxaas-azure/known-issues.md" >}}) page.
+To see a list of currently active issues, visit the [Known issues]({{< ref "/nginxaas-azure/known-issues" >}}) page.
 
 To review older entries, visit the [Changelog archive]({{< ref "/nginxaas-azure/changelog/archive" >}}) section.
 
+## July 29, 2026
+  NGINX Plus 37.0.4 is now available on the **Preview** [upgrade channel]({{< ref "/nginxaas-azure/quickstart/upgrade-channels" >}}) and will be promoted to the **Stable** upgrade channel starting the week of August 17, 2026.
+
+  NGINX Plus 37.0 introduced [several changes](https://docs.nginx.com/nginx/releases/#pls.37.0.4) including new default settings that can affect the behavior of your deployment and it is imperative that you take one of the following steps before your deployment is upgraded:
+- Test your configuration on the Preview channel to confirm the new changes do not adversely affect the operation of your applications.
+- Preserve your R36 behavior by following the recommendations in the [NGINX Plus 37.0 behavior impact]({{< ref "/nginxaas-azure/known-issues/#nginx-plus-370-behavior-impact" >}}) section of the **Known issues** page.
+
 ## July 2, 2026
 
   NGINX Plus 37.0 (PLS.37.0) introduced [updated defaults](https://docs.nginx.com/nginx/releases/#r37.0) that affected some users. Deployments on the **Stable** [Upgrade Channel]({{< ref "/nginxaas-azure/quickstart/upgrade-channels.md" >}}) have been reverted to NGINX Plus Release 36 (R36) while we work with affected users to resolve concerns.

content/nginxaas-azure/known-issues.md

@@ -11,6 +11,50 @@ url: /nginxaas-azure/known-issues/
 
 List of known issues in the latest releases of F5 NGINXaaS for Azure.
 
+### NGINX Plus 37.0 behavior impact
+
+The release of NGINX Plus [37.0]({{< ref "/nginx/releases/#pls.37.0.4" >}}) introduces the following behavioral changes that may affect upstream applications:
+- HTTP/1.1 is now the default protocol for connecting to proxy or upstream servers.
+- Keepalive connections between NGINX and upstream servers are enabled by default.
+- Upstream shared memory zone requires an additional 1KB of memory per upstream server.
+
+To preserve existing behavior, make the following configuration changes to your NGINX Plus [R36-P8]({{< ref "/nginx/releases/#r36" >}}) configuration before NGINXaaS is upgraded to NGINX Plus 37.0.4 on the stable upgrade channel (August 17, 2026):
+
+- Explicitly set HTTP/1.0 as the default protocol for communicating with upstream or proxy servers:
+  ```shell
+  http {
+     proxy_http_version 1.0;
+     ...
+  }
+  ```
+
+- For all upstream or proxy servers that explicitly use the HTTP/1.1 protocol, use the [proxy_set_header](https://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_set_header) directive to ensure the connection is closed after completing the request-response cycle:
+  ```shell
+  location  /example {
+     proxy_set_header Connection "close";
+     proxy_http_version 1.1;
+     ...
+  }
+  ```
+
+- Set the [`keepalive`](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#keepalive) directive to zero for all HTTP/1.1 upstream blocks where it has not been explicitly defined:
+  ```shell
+  upstream backend {
+     ...
+     keepalive 0;
+  }
+  ```
+
+- Increase the upstream block shared memory [zone](https://nginx.org/en/docs/http/ngx_http_upstream_module.html#zone) size by approximately 1KB per upstream server, rounding up to leave some headroom:
+  ```shell
+  upstream backend {
+     ...
+     zone backend 64k; # Example only. New size depends on previous settings.
+  }
+  ```
+
+For further information on the new behavior of NGINX Plus 37.0, see **NGINX Plus PLS.37.0.4.1 LTS** [Upgrade Notes]({{< ref "/nginx/releases/#pls.37.0.4" >}}).
+
 ### {{% icon-bug %}} Terraform fails to apply due to validation errors, but creates "Failed" resources in Azure (ID-4424)
 
 Some validation errors are caught later in the creation process, and can leave behind "Failed" resources in Azure. An example initial failure might look like: