docs: add log_collection_for_troubleshooting

src/app/_meta.global.ts

@@ -47,6 +47,11 @@ export default {
           node_pod_evictor: {}
         }
       },
+      troubleshooting: {
+        items: {
+          log_collection_for_troubleshooting: {}
+        }
+      },
       tips: {
         items: {
           ecr_auto_create: {},

src/content/guide/troubleshooting/log_collection_for_troubleshooting.mdx

@@ -0,0 +1,34 @@
+---
+title: Log Collection for Troubleshooting
+---
+
+# Log Collection for Troubleshooting
+This document provides a procedure to collect CloudPilot AI component logs from a Kubernetes cluster. The logs contain detailed operational information for troubleshooting CloudPilot AI-specific issues, such as why nodes are provisioned or de-provisioned. These logs do not include any sensitive system information.
+
+## Export Collected Log Steps
+
+1. **Connect to the target Kubernetes cluster** with administrative privileges.
+2. **Export collected logs for sharing**:
+   ```bash
+   kubectl -n cloudpilot port-forward svc/cloudpilot-workload-log-collector-svc 8080:8080 & PID=$!; timeout=10; found=0; while [ $timeout -gt 0 ]; do if nc -z -w 1 localhost 8080 2>/dev/null; then found=1; break; fi; sleep 1; ((timeout--)); done; [ $found -eq 1 ] && curl -s http://localhost:8080/logs > "logs_$(date +%Y%m%d%H%M%S).tgz" || echo "Timeout"; kill $PID 2>/dev/null; wait $PID 2>/dev/null
+   ```
+   **Command breakdown**:
+   - Establishes a port-forward tunnel to the `cloudpilot-workload-log-collector-svc` service.
+   - Waits up to 10 seconds for the service endpoint to become reachable.
+   - Uses `curl` to fetch logs via the exposed HTTP endpoint.
+   - Saves logs as a timestamped `.tgz` file (e.g., `logs_20231015143045.tgz`).
+   - Gracefully terminates the background port-forward process.
+
+   After running this command, you will get one tgz file containing the detailed logs (e.g., logs_20231015143045.tgz). You can now share it with the CloudPilot AI Support team for any further questions.
+
+If you encounter issues while exporting logs, please check the following:
+- Your Kubernetes context is set to the correct cluster.
+- The `cloudpilot-workload-log-collector-svc` service exists (`kubectl -n cloudpilot get svc`).
+- Your permissions allow port-forwarding and pod access in the `cloudpilot` namespace.
+
+## Security Considerations
+
+The logs contain operational metadata (e.g., pod statuses, API call traces) but exclude:
+- SSensitive workload data, such as pod environment variables
+- Secrets or credentials
+- Network traffic payloads