fix(vllm): use get_latest_gh_tag, lower default resources to 2CPU/4GB + add model requirements note

Author: Rohit (MC Backend Agent)

ct/vllm.sh

@@ -30,7 +30,7 @@ function update_script() {
     exit
   fi
 
-  RELEASE=$(curl -fsSL https://api.github.com/repos/vllm-project/vllm/releases/latest | grep "tag_name" | awk -F '"' '{print $4}')
+  RELEASE=$(get_latest_gh_tag "vllm-project/vllm")
   if [[ ! -f /opt/vLLM_version.txt ]] || [[ "${RELEASE}" != "$(cat /opt/vLLM_version.txt)" ]]; then
     if [[ ! -f /opt/vLLM_version.txt ]]; then
       touch /opt/vLLM_version.txt

install/vllm-install.sh

@@ -30,7 +30,7 @@ msg_info "Setting up Python Environment"
 $STD uv venv /opt/vllm/.venv
 msg_ok "Set up Python Environment"
 
-RELEASE=$(curl -fsSL https://api.github.com/repos/vllm-project/vllm/releases/latest | grep "tag_name" | awk -F '"' '{print $4}')
+RELEASE=$(get_latest_gh_tag "vllm-project/vllm")
 VLLM_VERSION="${RELEASE#v}"
 
 msg_info "Installing ${APP} ${RELEASE} (Patience — this takes 5-15 minutes)"

json/vllm.json

@@ -17,8 +17,8 @@
       "type": "default",
       "script": "ct/vllm.sh",
       "resources": {
-        "cpu": 8,
-        "ram": 16384,
+        "cpu": 2,
+        "ram": 4096,
         "hdd": 40,
         "os": "Ubuntu",
         "version": "24.04"
@@ -33,6 +33,10 @@
     {
       "text": "GPU passthrough requires a privileged LXC container with NVIDIA drivers installed on the Proxmox host. Edit /etc/vllm/vllm.env to configure model and settings before starting the service.",
       "type": "info"
+    },
+    {
+      "text": "Default resources (2 CPU, 4GB RAM) are sufficient for installation only. Running models requires significantly more: small models (1-3B) need 4+ CPU and 8GB+ RAM; large models (7B+) require a GPU with 8GB+ VRAM and 16GB+ RAM.",
+      "type": "warning"
     }
   ]
 }