Merge pull request #5 from svc-design/te0hq5-codex/使用ansible安装k8s并配置gpu驱动

Refine gpu-k8s role variables

docs/gpu-k8s-role.md

@@ -10,7 +10,8 @@ The role performs three main tasks:
 2. **Install NVIDIA drivers and container runtime** on the target hosts so that Kubernetes can access GPU resources.
 3. **Verify GPU access** by deploying the official NVIDIA device plugin and running a small CUDA workload.
 
-The following command is used to create the cluster:
+
+The following command is used to create the cluster (example with one master and one worker):
 
 ```bash
 sealos run \
@@ -35,8 +36,27 @@ Add the role to your playbook:
     - gpu-k8s
 ```
 
+
+Example playbook snippet defining the IP lists:
+
+```yaml
+- hosts: all
+  vars:
+    master_ips:
+      - "172.16.11.120"
+    node_ips:
+      - "172.16.11.152"
+  roles:
+    - gpu-k8s
+```
+
+The playbook expects `master_ips` and `node_ips` variables which are lists of IP addresses. Up to
+three masters can be specified.
+
+
 Run the playbook with your inventory that contains the master and node IP addresses.
 
+
 ```bash
 ansible-playbook -i inventory/hosts/all playbooks/demo_gpu_k8s.yml
 ```

playbooks/demo_gpu_k8s.yml

@@ -1,4 +1,9 @@
 - hosts: all
   become: true
+  vars:
+    master_ips:
+      - "172.16.11.120"
+    node_ips:
+      - "172.16.11.152"
   roles:
     - gpu-k8s

playbooks/roles/vhosts/gpu-k8s/defaults/main.yml

@@ -2,7 +2,7 @@
 sealos_version: v1.29.9
 cilium_version: v1.13.4
 helm_version: v3.9.4
-master_ip: "172.16.11.120"
-node_ip: "172.16.11.152"
+master_ips: []  # List of up to three master node IPs
+node_ips: []    # List of worker node IPs
 sealos_cmd_env: '{}'
 kubeadm_init_cmd: "kubeadm init --skip-phases=addon/kube-proxy"

playbooks/roles/vhosts/gpu-k8s/tasks/install_cluster.yml

@@ -4,10 +4,10 @@
       registry.cn-shanghai.aliyuncs.com/labring/kubernetes:{{ sealos_version }} \
       registry.cn-shanghai.aliyuncs.com/labring/cilium:{{ cilium_version }} \
       registry.cn-shanghai.aliyuncs.com/labring/helm:{{ helm_version }} \
-      --masters {{ master_ip }} \
-      --nodes {{ node_ip }} \
+      --masters {{ master_ips | join(',') }} \
+      --nodes {{ node_ips | join(',') }} \
       --env '{{ sealos_cmd_env }}' \
       --cmd "{{ kubeadm_init_cmd }}"
   args:
     executable: /bin/bash
-  when: inventory_hostname == master_ip
+  when: inventory_hostname == (master_ips | first)

playbooks/roles/vhosts/gpu-k8s/tasks/run_test.yml

@@ -2,7 +2,7 @@
   shell: kubectl apply -f https://raw.githubusercontent.com/NVIDIA/k8s-device-plugin/v0.14.5/nvidia-device-plugin.yml
   args:
     executable: /bin/bash
-  when: inventory_hostname == master_ip
+  when: inventory_hostname == (master_ips | first)
 
 - name: Run CUDA validation pod
   shell: |
@@ -10,4 +10,5 @@
     kubectl delete pod gpu-test --wait
   args:
     executable: /bin/bash
-  when: inventory_hostname == master_ip
+  when: inventory_hostname == (master_ips | first)
+