Common node fault tunables

Fault tunables which are common to all the node faults are described here. These tunables can be provided at .spec.experiment[*].spec.components.env in the chaosengine.

Target single node

It defines the name of the target node subject to chaos. You can tune it using the TARGET_NODE environment variable. It contains a single node name.

Note

It supports node drain, node taint, node restart, kubelet service kill, and docker service kill faults.

Use the following example to tune it:

## provide the target node name
## it is applicable for the [node-drain, node-taint, node-restart, kubelet-service-kill, docker-service-kill]
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  chaosServiceAccount: litmus-admin
  experiments:
  - name: node-drain
    spec:
      components:
        env:
        # name of the target node
        - name: TARGET_NODE
          value: 'node01'

Target multiple nodes

It defines the comma-separated names of the target nodes subject to chaos. You can tune it using the TARGET_NODES environment variable. NOTE: It supports node CPU hog, node memory hog, and node I/O stress faults.

Use the following example to tune it:

## provide the comma separated target node names
## it is applicable for the [node-cpu-hog, node-memory-hog, node-io-stress]
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  chaosServiceAccount: litmus-admin
  experiments:
  - name: node-cpu-hog
    spec:
      components:
        env:
        # comma separated target node names
        - name: TARGET_NODES
          value: 'node01,node02'

Target nodes with labels

It defines the labels of the target node(s) subject to chaos. You can tune it using the NODE_LABEL environment variable. It is mutually exclusive with the TARGET_NODE environment variable. If TARGET_NODE environment variable is set, the nodes provided will be used. Otherwise, it derives the node name(s) by matching it with the node labels.

Use the following example to tune it:

## provide the labels of the targeted nodes
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  chaosServiceAccount: litmus-admin
  experiments:
  - name: node-cpu-hog
    spec:
      components:
        env:
        # labels of the targeted node
        # it will derive the target nodes if TARGET_NODE(S) ENV is not set
        - name: NODE_LABEL
          value: 'key=value'

Node affected percentage

It defines the percentage of nodes subject to chaos by matching the node labels. You can tune it using the NODES_AFFECTED_PERC environment variable. If NODES_AFFECTED_PERC environment variable is set to empty or 0, it targets a minimum of one node. It supports node CPU hog, node memory hog, and node I/O stress faults.

Use the following example to tune it:

## provide the percentage of nodes to be targeted with matching labels
## it is applicable for the [node-cpu-hog, node-memory-hog, node-io-stress]
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  chaosServiceAccount: litmus-admin
  experiments:
  - name: node-cpu-hog
    spec:
      components:
        env:
        # percentage of nodes to be targeted with matching node labels
        - name: NODES_AFFECTED_PERC
          value: '100'
        # labels of the targeted node
        # it will derive the target nodes if TARGET_NODE(S) ENV is not set
        - name: NODE_LABEL
          value: 'key=value'