add release note for v1.6 (#178)

add release note for v1.6

Author: hippogr

docs/LuciaTrainingPlatform/blog/2026-04-30-release-1-6.md

@@ -0,0 +1,58 @@
+---
+slug: release-ltp-v1.6
+title: Releasing Lucia Training Platform v1.6
+author: Lucia Training Platform Team
+tags: [ltp, announcement, release]
+---
+
+We are pleased to announce the official release of **Lucia Training Platform v1.6.0**!
+
+## Lucia Training Platform v1.6.0 Release Notes
+
+This release focuses on security hardening, Docker image optimization, infrastructure upgrades, and bug fixes across the platform.
+
+## Platform Features & Bug Fixes
+- Upgraded webportal to Node.js 24 and removed the separate webportal-dind service — webportal now runs directly without Docker-in-Docker, simplifying deployment and reducing image size
+- Fixed job-detail page error handling for permission denied errors — now shows a clear message instead of infinite loading
+- Fixed job YAML and output log display issues on the webportal
+- Added support for tagging different types of GPUs
+- Skipped validation job submission for CPU nodes
+- Made Prometheus retention size configurable per service to prevent disk full issues
+- Added tool to preserve application tokens when revoking all tokens
+- Removed cronjob of abnormal-detector when stopping the service
+- Fixed exception when no name exists in filter
+
+## Docker Image Optimization
+- Reduced Docker image sizes for cluster-local-storage, copilot-chat, dashboard-data-backup, utilization-reporter, abnormal-detector, cert-expiration-checker, cluster-utilization, reverse-proxy, and model-proxy
+- Upgraded metrics-cleaner base image from Python 3.7 to 3.12-slim
+- Cleaned up job-exporter Docker image
+
+## Infrastructure & Networking
+- Updated Cilium from 1.18.6 to 1.18.9
+- Updated Go version to 1.25 across all Go-based components
+- Homebrew build for kube-scheduler and Grafana container images
+- Downgraded kube-scheduler version to match service Kubernetes version
+- Added IPoIB subnet route in init.sh to fix InfiniBand TCP connectivity on NetworkManager-managed nodes
+- Fixed DNS problem for cluster-local-storage
+- Fixed zlib 1.3.1 missing issue for pylon
+- Added Managed Identity support for build scripts
+- Made imagePullSecrets conditional to eliminate FailedToRetrieveImagePullSecret warnings
+- Removed secret deployment for image pull in favor of ACR credentials
+
+## Alert Manager & Node Management
+- Fixed KeyError when alert-parser processes validating nodes with no alerts
+- Downgraded hardware issues without Azure FaultCode to triaged_unknown to avoid broken OFR pipeline
+- Prevented node-recycler from submitting duplicate OFR tickets for the same node
+- Skipped classification for cordoned nodes with empty NodeId to prevent OFR pipeline stalling
+
+## Security
+- Updated Go toolchain and packages across all Go-based services
+- Updated Node.js packages for rest-server, alert-handler, job-status-change-notification, database-controller, and webportal
+- Updated Python packages for copilot-chat
+- Fixed S360 vulnerabilities across 13 container images including openssl, axios, follow-redirects, lodash, nodemailer, and minimatch
+- Updated go-ntlmssp to 0.1.1 for reverse proxy
+- Updated k8s-rdma-shared-dev-plugin to adapt to latest gRPC package
+
+## CI/CD
+- Updated CI workflow to filter dev-box from changed services detection
+- Removed all existing statefulsets in the system during cleanup instead of only config-defined ones