fix error handling in job to remove orphaned AKS clusters, update README

README.md

@@ -155,8 +155,9 @@ This requires the following:
    Create the app in "My Organization Only". It does not need a redirect URI or any other optional settings.
    Create a client secret for this app. If you want the client secret to have a long lifetime, you can use the `az` CLI to create it.
    In the Subscription's IAM settings, assign this app the role "Azure Kubernetes Service Contributor Role" to allow
-   the app to manage AKS clusters. Do not grant this app permissions in any other Subscription or use it for any
-   other purpose.
+   the app to manage AKS clusters. Also assign this app the role "Reader" to allow it to read all resources
+   (used by the `remove-orphaned-aks-clusters` CI task).
+   Do not grant this app permissions in any other Subscription or use it for any other purpose.
 3. Configure the pipelines with the app's Application (client) ID, Client Secret, and Directory (tenant) ID
    as the appropriate secret values.
 

pipelines/shared-tasks/remove-orphaned-aks-clusters/task.sh

@@ -44,9 +44,15 @@ cat all-clusters.txt
 echo
 
 # Remove clusters with unexpected name formats. They might have been created manually for testing.
-cat all-clusters.txt | grep -E '^aks-[a-f0-9]+ ' >ci-clusters.txt
+# The curly braces and "or true" syntax prevents errors when the grep fails to find any matches.
+cat all-clusters.txt | { grep -E '^aks-[a-f0-9]+ ' || true; } >ci-clusters.txt
 
-echo "Only those clusters with expected naming convention:"
+if [[ ! -s ci-clusters.txt ]]; then
+  echo "No clusters matching the expected naming convention were found."
+  exit 0
+fi
+
+echo "Only those clusters matched the expected naming convention:"
 cat ci-clusters.txt
 echo