Troubleshooting

Security risk error

Alan AI Studio currently uses self-signed certificates that are available in the cfg/ans-development/certs folder of the installation package. Browsers may warn against using self-signed certificates.

To avoid it, add an exception in Chrome:

1. In the top-right corner of the Chrome window, click three dots and select Settings.

2. Scroll down and click Advanced to expand the advanced settings.

3. Under the Privacy and Security section, click Manage Certificates.

4. In the Certificates window, go to the Authorities tab.

5. Click Import and select the self-signed certificate file in the installation package.

Alan AI Cloud deployment works incorrectly

You may experience the following symptoms:

• Unable to access Alan Studio through a web browser

• Internal errors or general errors are displayed as toaster notifications in the top right corner of Alan AI Studio

• The Alan AI button is unable to connect (the loader on the AI agent button keeps spinning indefinitely)

• The Disconnected banner appearing in Alan AI Studio

• Infinite loading animation in Alan Studio Debug Chat (disconnected or model is updating)

Steps to troubleshoot:

1. In the Terminal, run:

   Terminal

   kubectl get pods
   

2. Check the output to ensure all pods are in the 1/1 Running state.

3. All pods, except for alan-dialog pods, should be singular:

   1. If any other pod is duplicated, delete the oldest duplicate using the following command:

      Terminal

      kubectl delete pod {pod_name}
      

   2. Check the pods status using the following command:

      Terminal

      kubectl get pods
      

   3. If the pod deleted at the previous step reappears and duplicates again, run the following commands in sequence:

      Terminal

      ./scripts/local-helm-uninstall.sh
      
      ./scripts/local-delete-alan-pods.sh
      
      ./scripts/local-helm-install.sh
      

   4. All the data like projects, statistics and project settings are stored in persistent volumes and will remain unaffected by the reinstallation.

4. If the number of pods is correct but some aren’t in the 1/1 Running state, use the following command to check the pod logs:

   Terminal

   kubectl describe pod {pod_name}
   

   1. Common issues found in the logs may include:

      • Unbound Persistent Volume claim: indicates Kubernetes and Docker can’t allocate requested resources, possibly due to Docker errors or misconfigurations.

      • Unsupported architecture manifest

   2. Recommended configuration for Docker:

      • CPU limit: > 4

      • Memory limit: > 12 GB

      • Swap: > 2 GB

      • Virtual disk limit: > 100 GB

   3. Try restarting Docker. If this doesn’t help, restart your workstation to free up the necessary resources needed for local deployment.

   4. If the deployment still doesn’t work after a restart, reinstall it by running the following commands in sequence:

      Terminal

      ./scripts/local-helm-uninstall.sh
      
      ./scripts/local-delete-alan-pods.sh
      
      ./scripts/local-helm-install.sh