--- id: pod-memory-hog title: Pod Memory Hog Details sidebar_label: Pod Memory Hog original_id: pod-memory-hog --- --- ## Experiment Metadata
Type Description Tested K8s Platform
Generic Consume memory resources on the application container GKE, Packet(Kubeadm), Minikube, EKS
## Prerequisites - Ensure that the Litmus Chaos Operator is running by executing `kubectl get pods` in operator namespace (typically, `litmus`). If not, install from [here](https://docs.litmuschaos.io/docs/getstarted/#install-litmus) - 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](https://hub.litmuschaos.io/api/chaos/1.4.1?file=charts/generic/pod-memory-hog/experiment.yaml) - Cluster must run docker container runtime ## Entry Criteria - Application pods are healthy on the respective nodes before chaos injection ## Exit Criteria - Application pods are healthy on the respective nodes post chaos injection ## Details - 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. ## Integrations - Pod Memory can be effected using the chaos library: `litmus` ## Steps to Execute the Chaos Experiment - This Chaos Experiment can be triggered by creating a ChaosEngine resource on the cluster. To understand the values to provide in a ChaosEngine specification, refer [Getting Started](getstarted.md/#prepare-chaosengine) - Follow the steps in the sections below to create the chaosServiceAccount, prepare the ChaosEngine & execute the experiment. ### Prepare chaosServiceAccount 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. #### Sample Rbac Manifest [embedmd]: # "https://raw.githubusercontent.com/litmuschaos/chaos-charts/master/charts/generic/pod-memory-hog/rbac.yaml yaml" ```yaml --- apiVersion: v1 kind: ServiceAccount metadata: name: pod-memory-hog-sa namespace: default labels: name: pod-memory-hog-sa --- apiVersion: rbac.authorization.k8s.io/v1beta1 kind: Role metadata: name: pod-memory-hog-sa namespace: default labels: name: pod-memory-hog-sa rules: - apiGroups: ["", "litmuschaos.io", "batch"] resources: [ "pods", "jobs", "events", "pods/log", "chaosengines", "chaosexperiments", "chaosresults", ] verbs: ["create", "list", "get", "patch", "update", "delete"] --- apiVersion: rbac.authorization.k8s.io/v1beta1 kind: RoleBinding metadata: name: pod-memory-hog-sa namespace: default labels: name: pod-memory-hog-sa roleRef: apiGroup: rbac.authorization.k8s.io kind: Role name: pod-memory-hog-sa subjects: - kind: ServiceAccount name: pod-memory-hog-sa namespace: default ``` ### Prepare ChaosEngine - Provide the application info in `spec.appinfo` - Provide the auxiliary applications info (ns & labels) in `spec.auxiliaryAppInfo` - Override the experiment tunables if desired in `experiments.spec.components.env` - To understand the values to provided in a ChaosEngine specification, refer [ChaosEngine Concepts](chaosengine-concepts.md) #### Supported Experiment Tunables
Variables Description Type Notes
TARGET_CONTAINER Name of the container subjected to Memory stress Mandatory
MEMORY_CONSUMPTION The amount of memory used of hogging a Kubernetes pod (megabytes) Optional Defaults to 500MB
TOTAL_CHAOS_DURATION The time duration for chaos insertion (seconds) Optional Defaults to 60s
LIB_IMAGE The image used by the litmus (only supported) lib Optional Defaults to `litmuschaos/app-memory-stress:latest`
RAMP_TIME Period to wait before and after injection of chaos in sec Optional
INSTANCE_ID A user-defined string that holds metadata/info about current run/instance of chaos. Ex: 04-05-2020-9-00. This string is appended as suffix in the chaosresult CR name. Optional Ensure that the overall length of the chaosresult CR is still < 64 characters
#### Sample ChaosEngine Manifest [embedmd]: # "https://raw.githubusercontent.com/litmuschaos/chaos-charts/master/charts/generic/pod-memory-hog/engine.yaml yaml" ```yaml apiVersion: litmuschaos.io/v1alpha1 kind: ChaosEngine metadata: name: nginx-chaos namespace: default spec: # It can be true/false annotationCheck: "true" # It can be active/stop engineState: "active" #ex. values: ns1:name=percona,ns2:run=nginx auxiliaryAppInfo: "" appinfo: appns: "default" applabel: "app=nginx" appkind: "deployment" chaosServiceAccount: pod-memory-hog-sa monitoring: false # It can be delete/retain jobCleanUpPolicy: "delete" experiments: - name: pod-memory-hog spec: components: env: # Provide name of target container # where chaos has to be injected - name: TARGET_CONTAINER value: "nginx" # Enter the amount of memory in megabytes to be consumed by the application pod # default: 500 (Megabytes) - name: MEMORY_CONSUMPTION value: "" - name: TOTAL_CHAOS_DURATION value: "60" # in seconds ``` ### Create the ChaosEngine Resource - Create the ChaosEngine manifest prepared in the previous step to trigger the Chaos. `kubectl apply -f chaosengine.yml` ### Watch Chaos progress - Set up a watch on the applications interacting/dependent on the affected pods and verify whether they are running `watch kubectl get pods -n ` ### Check Chaos Experiment Result - Check whether the application stack is resilient to Memory spikes on the app replica, once the experiment (job) is completed. The ChaosResult resource name is derived like this: `-`. `kubectl describe chaosresult nginx-chaos-pod-memory-hog -n ` ## Pod Memory Hog Experiment Demo - A sample recording of this experiment will be added very soon.