Run an Agent on Every Node

The platform team needs the log-agent collector on every machine in the cluster, including the control plane and any node that a team reserves with its own taint, so that no node goes unmonitored. Nodes added or tainted later must keep getting it automatically.

Task

• Create a DaemonSet log-agent in namespace analytics running the image alpine:3.20 with the command sleep infinity.
• Its Pods carry the label app=log-agent.
• Exactly one log-agent Pod runs and is ready on each of the three nodes: controlplane, node01 and node02.
• The agent tolerates every taint (any key, any value, any effect), so it keeps running on nodes that teams taint later, for example dedicated=gpu:NoSchedule. Tolerating only the control-plane taint is not enough.
• Do not remove or change the taint on controlplane.

A practice cluster is running on this machine: start with kubectl get nodes and kubectl describe node controlplane.