Description

Solutions based on AKS 1.25.* cluster version with Nginx ingress controller version 1.3.0 might face issues and cause downtime. The root cause is that  AKS 1.25.* is not compatible with Nginx ingress controller 1.3.0.

Solution

Important note:

Before you perform the below steps, double check your Nginx ingress controller version. In your AKS, check the Nginx pod "nginx-ingress-xxx-xxxx". You can find the version in the YAML file like below.

In case you have any questions or need any further assistance, raise a support case from the Sitecore Support Portal.

 

 

 

 

As a temporary solution, consider increasing resources of Nginx twice (x2) in Application Repo -> /config/resources/resources.json:

 

Planned further actions for Customers

Sitecore continues to work on solving the issue and will provide an update later about the final solution. Follow the updates in this article.

• UPDATE 1: The fix for this issue (upgrade to NGINX 1.8.0) will be included in the future release.

  As a temporary solution, perform the following steps:

  1. In your 'Application' repository find the requirements.yaml file.
  2. In the mentioned file replace the package name with the sitecore-managedcloud-0.0.332472.tar.gz
     For example, replace:
     

     sitecore/managedcloud/sitecore-managedcloud-0.0.318880.tar.gz

     with:

     sitecore/managedcloud/sitecore-managedcloud-0.0.332472.tar.gz

  3. Merge the change to the "main" branch.
  4. Run the Application Pipeline.

  After the Application pipeline finishes, all NGINX pods will be re-created with 1.8.0 version of the image.

• UPDATE 2: It was observed that the issue with OOM kill events on NGINX 1.8.0 is still present and pods are unstable due to insufficient memory, the previous(default) resource values for NGINX deployment should be increased as follows:

      "nginx": {
          "nginx": {
              "request_memory": "192Mi",
              "request_cpu": "128m",
              "limit_memory": "256Mi",
              "limit_cpu": "512m",
              "replicas": 2
          }

  Important note: Before making any changes, double check your Linux node resource utilization in default node pool, especially if you have custom workload which resides on Linux node.