Add or remove nodes
Or how to scale vertically. To replace the same node with a clean OS, remove it and add it again.
Add new nodes
Tip
You can add multiple nodes at the same time
Two places to edit —— 拓扑(名称、IP、磁盘、网卡)进 inventory,真实 MAC 进那个 不进版本控制的 secrets 文件(公开仓库里不要出现真实 MAC,#550):
metal/inventories/prod.yml
diff --git a/metal/inventories/prod.yml b/metal/inventories/prod.yml
index 7f6474a..1bb2cbc 100644
--- a/metal/inventories/prod.yml
+++ b/metal/inventories/prod.yml
@@ -8,3 +8,4 @@ metal:
workers:
hosts:
metal3: {ansible_host: 192.168.1.113, mac: "{{ metal_macs['metal3'] }}", disk: sda, network_interface: eno1}
+ metal4: {ansible_host: 192.168.1.114, mac: "{{ metal_macs['metal4'] }}", disk: sda, network_interface: eno1}
metal/inventories/group_vars/all/zz-secrets.yml
diff --git a/metal/inventories/group_vars/all/zz-secrets.yml b/metal/inventories/group_vars/all/zz-secrets.yml
--- a/metal/inventories/group_vars/all/zz-secrets.yml
+++ b/metal/inventories/group_vars/all/zz-secrets.yml
@@ -1,3 +1,4 @@
metal_macs:
metal3: "aa:bb:cc:00:00:03"
+ metal4: "aa:bb:cc:00:00:04"
这个文件被 gitignore,MAC 放哪里、真实值怎么取见 metal/README.md。
Install the OS and join the cluster:
make metal
That's it!
Remove a node
Danger
It is recommended to remove nodes one at a time
Remove it from the inventory:
metal/inventories/prod.yml
diff --git a/metal/inventories/prod.yml b/metal/inventories/prod.yml
index 7f6474a..d12b50a 100644
--- a/metal/inventories/prod.yml
+++ b/metal/inventories/prod.yml
@@ -4,7 +4,6 @@ metal:
hosts:
metal0: {ansible_host: 192.168.1.110, mac: "{{ metal_macs['metal0'] }}", disk: sda, network_interface: eno1}
metal1: {ansible_host: 192.168.1.111, mac: "{{ metal_macs['metal1'] }}", disk: sda, network_interface: eno1}
- metal2: {ansible_host: 192.168.1.112, mac: "{{ metal_macs['metal2'] }}", disk: sda, network_interface: eno1}
workers:
hosts:
metal3: {ansible_host: 192.168.1.113, mac: "{{ metal_macs['metal3'] }}", disk: sda, network_interface: eno1}
(它在 zz-secrets.yml 的 metal_macs 里的条目可以顺手删掉,留着不影响运行。)
Drain the node:
kubectl drain ${NODE_NAME} --delete-emptydir-data --ignore-daemonsets --force
Remove the node from the cluster
kubectl delete node ${NODE_NAME}
Shutdown the node:
ssh root@${NODE_IP} poweroff