Skip to main content

Common node fault tunables

Last updated on

These tunables apply to every node-level chaos fault in Harness Chaos Engineering. Set them on the fault in Chaos Studio to choose which nodes the experiment targets.


Target single node

Use the TARGET_NODE environment variable to name the single node the fault should run against.

TunableDescriptionDefault
TARGET_NODEName of the node to target.""

Supported by: Node drain, Node taint, Node restart, Kubelet service kill.

- name: TARGET_NODE
value: "node01"

Target multiple nodes

Use the TARGET_NODES environment variable to target a comma-separated list of nodes.

TunableDescriptionDefault
TARGET_NODESComma-separated list of node names to target.""

Supported by: Node CPU hog, Node memory hog, Node I/O stress, Node network loss, Node network latency.

- name: TARGET_NODES
value: "node01,node02"

Target nodes with labels

Use the NODE_LABEL environment variable to select target nodes by label. This is mutually exclusive with TARGET_NODE and TARGET_NODES. If both are set, the explicit node list wins.

TunableDescriptionDefault
NODE_LABELLabel selector in key=value form. The fault targets nodes whose labels match.""
- name: NODE_LABEL
value: "kubernetes.io/role=worker"

Node affected percentage

Use the NODES_AFFECTED_PERCENTAGE environment variable to target a percentage of nodes that match the label selector. A value of 0 (the default) means one node.

TunableDescriptionDefault
NODES_AFFECTED_PERCENTAGEPercentage of nodes (matching the selector) to target. 0 means one node.0

Supported by: Node CPU hog, Node memory hog, Node I/O stress.

- name: NODES_AFFECTED_PERCENTAGE
value: "100"
- name: NODE_LABEL
value: "kubernetes.io/role=worker"