MLflow wait_for: ignore_errors so install continues regardless

ansible/roles/base/tasks/main.yml

@@ -5,7 +5,7 @@
     update_cache: yes
     cache_valid_time: 3600
 
-- name: Upgrade all packages
+- name: Upgrade all packages (this takes 10–20 min on first run, please wait...)
   apt:
     upgrade: dist
     autoremove: yes
@@ -85,7 +85,7 @@
   args:
     creates: /opt/cezen/miniconda/envs/cezen
 
-- name: Install LangChain + LlamaIndex + HuggingFace in conda env
+- name: Install LangChain + LlamaIndex + HuggingFace in conda env (5–10 min, please wait...)
   become_user: cezen
   shell: |
     /opt/cezen/miniconda/bin/conda run -n cezen pip install \

ansible/roles/mlflow/tasks/main.yml

@@ -54,3 +54,4 @@
     host: localhost
     port: 5000
     timeout: 30
+  ignore_errors: true

install.sh

@@ -54,8 +54,9 @@ run_phase1() {
   echo "║   Cezen AI Suite — Phase 1: NVIDIA       ║"
   echo "╚══════════════════════════════════════════╝"
 
+  ANSIBLE_STDOUT_CALLBACK=yaml \
   ansible-playbook -i localhost, -c local "$ANSIBLE_DIR/phase1_nvidia.yml" \
-    -e "tier=$TIER" -v
+    -e "tier=$TIER"
 
   # Register phase 2 as a one-shot systemd service so it runs after reboot
   cat > /etc/systemd/system/cezen-phase2.service << EOF
@@ -100,8 +101,9 @@ run_phase2() {
     echo "✓ NVIDIA driver: $(nvidia-smi --query-gpu=driver_version --format=csv,noheader | head -1)"
   fi
 
+  ANSIBLE_STDOUT_CALLBACK=yaml \
   ansible-playbook -i localhost, -c local "$ANSIBLE_DIR/entry.yml" \
-    -e "tier=$TIER" -v
+    -e "tier=$TIER"
 
   # Disable one-shot service so it doesn't run again on next reboot
   systemctl disable cezen-phase2.service 2>/dev/null || true