Skip to content

Latest commit

 

History

History
86 lines (67 loc) · 2.17 KB

add-node.md

File metadata and controls

86 lines (67 loc) · 2.17 KB
Raw

Add a Cluster node

Adding a Kube Node to an existing cluster

1. Run cluster setup

From the dgx-setup repo

Edit the inventory file in the dgx-setup repo to add the new node(s)

  • New nodes should be defined with IP addresses under the [all] section and then added to the [node] section
ansible-playbook -i ./inventory -l lambda ./cluster-setup.yaml

2. Edit deepops Inventory

From the deepops repo

  • deepops/config/inventory Add new nodes to All with ip and ansible_host. Add names to kube-nodes

Add to all:vars

dploy_container_engine=False
download_run_once=True

3. Run Deepops expand

From the deepops repo

ansible-playbook -i ./config/inventory -l k8s-cluster ./submodules/kubespray/scale.yml

If sucessfull, check the node status with

kubectl get nodes

Issues

There may be some issues running the scale.yml playbook. Here are some common issues and resolutions.

Jinga Package version

Kubespray requires at least jinja 2.11. If you hit an error with setting up kubeadm do the following. Note this will uninstall the openshift client as well, so we are reinstalling it.

sudo yum remove python-jinja2
sudo pip uninstall ansible Jinja2
sudo pip install --upgrade ansible==2.9.5
sudo pip install --upgrade Jinja2==2.11.1
sudo pip install --upgrade setuptools
sudo pip install --upgrade openshift==0.11.2

See the following https://github.com/kubernetes-sigs/kubespray/issues/5958kub

4. Label Nodes

Label each node as a compute node.

kubectl label node NODE_NAME node-role.kubernetes.io/node=

Apply Custom node-type labels.

kubectl label node NODE_NAME system_type=lambda

Apply DCGM Exporter Label

kubectl label nodes <node-name> hardware-type=NVIDIAGPU

5. Validation

These are some steps to validate the cluster is working

Check System/Monitoring Pods

Validate that the system and monitoring pods are running on the new node

Run this for each new node to ensure all pods are running and not restarting

kubectl get pods -A -o wide |grep NODE_NAME

Check to make sure NFD annotations are working

Check to make sure we can see the Nvidia GPUs on each node

kubectl describe node NODE_NAME