Pod Memory Hog

Introduction¶

This experiment consumes the Memory resources on the application container on specified memory in megabytes.
It simulates conditions where app pods experience Memory spikes either due to expected/undesired processes thereby testing how the overall application stack behaves when this occurs.

Scenario: Stress the Memory

Pod Memory Hog

Uses¶

View the uses of the experiment

Memory usage within containers is subject to various constraints in Kubernetes. If the limits are specified in their spec, exceeding them can cause termination of the container (due to OOMKill of the primary process, often pid 1) - the restart of the container by kubelet, subject to the policy specified. For containers with no limits placed, the memory usage is uninhibited until such time as the Node level OOM Behaviour takes over. In this case, containers on the node can be killed based on their oom_score and the QoS class a given pod belongs to (bestEffort ones are first to be targeted). This eval is extended to all pods running on the node - thereby causing a bigger blast radius.

This experiment launches a stress process within the target container - which can cause either the primary process in the container to be resource constrained in cases where the limits are enforced OR eat up available system memory on the node in cases where the limits are not specified

Prerequisites¶

Verify the prerequisites

Ensure that Kubernetes Version > 1.16
Ensure that the Litmus Chaos Operator is running by executing kubectl get pods in operator namespace (typically, litmus).If not, install from here
Ensure that the pod-memory-hog experiment resource is available in the cluster by executing kubectl get chaosexperiments in the desired namespace. If not, install from here

Default Validations¶

View the default validations

The application pods should be in running state before and after chaos injection.

Minimal RBAC configuration example (optional)¶

NOTE

If you are using this experiment as part of a litmus workflow scheduled constructed & executed from chaos-center, then you may be making use of the litmus-admin RBAC, which is pre installed in the cluster as part of the agent setup.

View the Minimal RBAC permissions

---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: pod-memory-hog-sa
  namespace: default
  labels:
    name: pod-memory-hog-sa
    app.kubernetes.io/part-of: litmus
---
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
  name: pod-memory-hog-sa
  namespace: default
  labels:
    name: pod-memory-hog-sa
    app.kubernetes.io/part-of: litmus
rules:
  # Create and monitor the experiment & helper pods
  - apiGroups: [""]
    resources: ["pods"]
    verbs: ["create","delete","get","list","patch","update", "deletecollection"]
  # Performs CRUD operations on the events inside chaosengine and chaosresult
  - apiGroups: [""]
    resources: ["events"]
    verbs: ["create","get","list","patch","update"]
  # Fetch configmaps details and mount it to the experiment pod (if specified)
  - apiGroups: [""]
    resources: ["configmaps"]
    verbs: ["get","list",]
  # Track and get the runner, experiment, and helper pods log 
  - apiGroups: [""]
    resources: ["pods/log"]
    verbs: ["get","list","watch"]  
  # for creating and managing to execute comands inside target container
  - apiGroups: [""]
    resources: ["pods/exec"]
    verbs: ["get","list","create"]
  # deriving the parent/owner details of the pod(if parent is anyof {deployment, statefulset, daemonsets})
  - apiGroups: ["apps"]
    resources: ["deployments","statefulsets","replicasets", "daemonsets"]
    verbs: ["list","get"]
  # deriving the parent/owner details of the pod(if parent is deploymentConfig)  
  - apiGroups: ["apps.openshift.io"]
    resources: ["deploymentconfigs"]
    verbs: ["list","get"]
  # deriving the parent/owner details of the pod(if parent is deploymentConfig)
  - apiGroups: [""]
    resources: ["replicationcontrollers"]
    verbs: ["get","list"]
  # deriving the parent/owner details of the pod(if parent is argo-rollouts)
  - apiGroups: ["argoproj.io"]
    resources: ["rollouts"]
    verbs: ["list","get"]
  # for configuring and monitor the experiment job by the chaos-runner pod
  - apiGroups: ["batch"]
    resources: ["jobs"]
    verbs: ["create","list","get","delete","deletecollection"]
  # for creation, status polling and deletion of litmus chaos resources used within a chaos workflow
  - apiGroups: ["litmuschaos.io"]
    resources: ["chaosengines","chaosexperiments","chaosresults"]
    verbs: ["create","list","get","patch","update","delete"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
  name: pod-memory-hog-sa
  namespace: default
  labels:
    name: pod-memory-hog-sa
    app.kubernetes.io/part-of: litmus
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: pod-memory-hog-sa
subjects:
- kind: ServiceAccount
  name: pod-memory-hog-sa
  namespace: default

Use this sample RBAC manifest to create a chaosServiceAccount in the desired (app) namespace. This example consists of the minimum necessary role permissions to execute the experiment.

Experiment tunables¶

check the experiment tunables

Optional Fields

Variables	Description	Notes
MEMORY_CONSUMPTION	The amount of memory used of hogging a Kubernetes pod (megabytes)	Defaults to 500MB
NUMBER_OF_WORKERS	The number of workers used to run the stress process	Defaults to 1
TOTAL_CHAOS_DURATION	The time duration for chaos insertion (seconds)	Defaults to 60s
LIB	The chaos lib used to inject the chaos. Available libs are `litmus` and `pumba`	Defaults to `litmus`
LIB_IMAGE	Image used to run the helper pod.	Defaults to `litmuschaos/go-runner:1.13.8`
STRESS_IMAGE	Container run on the node at runtime by the pumba lib to inject stressors. Only used in LIB `pumba`	Default to `alexeiled/stress-ng:latest-ubuntu`
TARGET_PODS	Comma separated list of application pod name subjected to pod memory hog chaos	If not provided, it will select target pods randomly based on provided appLabels
TARGET_CONTAINER	Name of the target container under chaos.	If not provided, it will select the first container of the target pod
CONTAINER_RUNTIME	container runtime interface for the cluster	Defaults to containerd, supported values: docker, containerd and crio for litmus and only docker for pumba LIB
SOCKET_PATH	Path of the containerd/crio/docker socket file	Defaults to `/run/containerd/containerd.sock`
PODS_AFFECTED_PERC	The Percentage of total pods to target	Defaults to 0 (corresponds to 1 replica), provide numeric value only
RAMP_TIME	Period to wait before injection of chaos in sec
SEQUENCE	It defines sequence of chaos execution for multiple target pods	Default value: parallel. Supported: serial, parallel

Experiment Examples¶

Common and Pod specific tunables¶

Refer the common attributes and Pod specific tunable to tune the common tunables for all experiments and pod specific tunables.

Memory Consumption¶

It stresses the MEMORY_CONSUMPTION MB memory of the targeted pod for the TOTAL_CHAOS_DURATION duration.

Use the following example to tune this:

# define the memory consumption in MB
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  appinfo:
    appns: "default"
    applabel: "app=nginx"
    appkind: "deployment"
  chaosServiceAccount: pod-memory-hog-sa
  experiments:
  - name: pod-memory-hog
    spec:
      components:
        env:
        # memory consumption value
        - name: MEMORY_CONSUMPTION
          value: '500' #in MB
        - name: TOTAL_CHAOS_DURATION
          value: '60'

Workers For Stress¶

The worker's count for the stress can be tuned with NUMBER_OF_WORKERS ENV.

Use the following example to tune this:

# number of workers used for the stress
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  appinfo:
    appns: "default"
    applabel: "app=nginx"
    appkind: "deployment"
  chaosServiceAccount: pod-memory-hog-sa
  experiments:
  - name: pod-memory-hog
    spec:
      components:
        env:
        # number of workers for stress
        - name: NUMBER_OF_WORKERS
          value: '1'
        - name: TOTAL_CHAOS_DURATION
          value: '60'

Container Runtime Socket Path¶

It defines the CONTAINER_RUNTIME and SOCKET_PATH ENV to set the container runtime and socket file path.

CONTAINER_RUNTIME: It supports docker, containerd, and crio runtimes. The default value is docker.
SOCKET_PATH: It contains path of docker socket file by default(/run/containerd/containerd.sock). For other runtimes provide the appropriate path.

Pumba Chaos Library¶

It specifies the Pumba chaos library for the chaos injection. It can be tuned via LIB ENV. The defaults chaos library is litmus. Provide the stress image via STRESS_IMAGE ENV for the pumba library.

Use the following example to tune this:

# use the pumba lib for the memory stress
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: engine-nginx
spec:
  engineState: "active"
  annotationCheck: "false"
  appinfo:
    appns: "default"
    applabel: "app=nginx"
    appkind: "deployment"
  chaosServiceAccount: pod-memory-hog-sa
  experiments:
  - name: pod-memory-hog
    spec:
      components:
        env:
        # name of chaoslib
        # it supports litmus and pumba lib
        - name: LIB
          value: 'pumba'
        # stress image - applicable for pumba lib only
        - name: STRESS_IMAGE
          value: 'alexeiled/stress-ng:latest-ubuntu'
        - name: TOTAL_CHAOS_DURATION
          value: '60'