Windows EC2 memory hog

Last updated on Jun 5, 2026

Windows EC2 memory hog is an AWS chaos fault that allocates MEMORY_CONSUMPTION MB (or MEMORY_PERCENTAGE percent of total memory) on one or more Windows EC2 instances for a configurable duration. Targets are selected by EC2_INSTANCE_ID or EC2_INSTANCE_TAG. The fault dispatches a PowerShell allocation command via AWS Systems Manager Run Command and releases the memory at the end of the fault.

Use this fault to test how a Windows-hosted workload behaves under sustained memory pressure: whether the application sheds load when memory is tight, whether the working set spills to swap (paging file), and whether memory-based monitoring catches the pressure quickly.

Run your first experiment

If you have not configured the chaos infrastructure yet, go to Quickstart to install the chaos infrastructure and run an experiment end to end.

Use cases

Run this fault when you want to answer concrete questions like:

Memory pressure handling: When free memory shrinks, does the application sustain throughput, or does paging cause throughput collapse?
Service recovery: Do Windows services (IIS, SQL Server Express, application services) recover cleanly when memory is freed?
Autoscaling: Does a memory-based scaling policy fire and add capacity?
Monitoring SLA: Does the alert on Memory % Committed Bytes In Use fire within the SLA?
Recovery time: When the chaos ends, how quickly does memory return to baseline?

Prerequisites

Kubernetes version: 1.21 or later for the chaos infrastructure cluster. Go to What's supported to confirm distribution support.
Target Windows EC2 instances: Each instance is in the running state and registered with AWS Systems Manager.
Selection criteria: Either EC2_INSTANCE_ID is set or EC2_INSTANCE_TAG is set.
AWS credentials available: Either an AWS credentials file uploaded as a File Secret in Harness Secret Manager (see Authentication below) or an IAM role for service accounts (IRSA) bound to the chaos infrastructure service account.
IAM permissions granted: The credentials or role include the permissions listed below.

Supported environments

Platform	Support status
Windows Server 2016 / 2019 / 2022 EC2	Supported
Linux EC2 instances	Not supported (use EC2 memory hog)
AWS regions	Supported in every commercial region; pass the region in `REGION`

Permissions required

The IAM principal that the chaos pod uses (the credentials mounted from the Harness Secret Manager file secret, the IRSA role on the chaos service account, or the role assumed via ASSUME_ROLE_ARN) needs the following AWS actions.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "ec2:DescribeInstances",
        "ec2:DescribeInstanceStatus"
      ],
      "Resource": "*"
    },
    {
      "Effect": "Allow",
      "Action": [
        "ssm:SendCommand",
        "ssm:CancelCommand",
        "ssm:GetCommandInvocation",
        "ssm:DescribeInstanceInformation"
      ],
      "Resource": "*"
    },
    {
      "Effect": "Allow",
      "Action": [
        "ec2messages:AcknowledgeMessage",
        "ec2messages:DeleteMessage",
        "ec2messages:FailMessage",
        "ec2messages:GetEndpoint",
        "ec2messages:GetMessages",
        "ec2messages:SendReply"
      ],
      "Resource": "*"
    }
  ]
}

Go to common policy for all AWS faults to use a single superset IAM policy across every AWS fault.

Authentication

The fault supports three credential delivery models. Pick one based on how your chaos infrastructure is deployed.

Method	When to use it	How to configure
Harness Secret Manager file secret	Chaos infrastructure runs outside EKS, or you want explicit static credentials	Upload the AWS credentials file as a File Secret in Harness Secret Manager and reference its identifier via `AWS_AUTHENTICATION_SECRET`
IAM Roles for Service Accounts (IRSA)	Chaos infrastructure runs in EKS and uses an OIDC-bound service account	No tunable changes; the chaos pod inherits the role automatically. Go to AWS IAM integration to set it up
Assume role	The fault needs to act in a different account or with elevated permissions	Set `ASSUME_ROLE_ARN` to the role ARN; the chaos pod assumes the role on top of its base credentials

When using the Harness Secret Manager method, the contents of the File Secret should be the AWS credentials file in the standard ~/.aws/credentials format:

[default]
aws_access_key_id = REPLACE_WITH_ACCESS_KEY_ID
aws_secret_access_key = REPLACE_WITH_SECRET_ACCESS_KEY

Upload this file as a File Secret in Harness Secret Manager (Project Setup → Secrets → New File Secret), and pass the secret identifier in AWS_AUTHENTICATION_SECRET when configuring the fault.

Go to AWS named profile for chaos to switch between profiles inside a single credentials file.

Fault tunables

Configure the following fault parameters when you add Windows EC2 memory hog to an experiment in Chaos Studio. Defaults are shown for reference.

Required parameters

Tunable	Description	Default
`REGION`	AWS region that hosts the target instances (for example `us-east-1`).	(required)

Targeting parameters

Tunable	Description	Default
`EC2_INSTANCE_ID`	Comma-separated list of EC2 instance IDs to affect.	`""`
`EC2_INSTANCE_TAG`	Tag in `Key:Value` form matching candidate instances when `EC2_INSTANCE_ID` is empty.	`""`
`INSTANCE_AFFECTED_PERC`	Percentage of matching instances (by tag) to affect. `0` corresponds to one instance.	`0`

Chaos parameters

Tunable	Description	Default
`MEMORY_CONSUMPTION`	Memory to allocate in MB. Set to `0` and use `MEMORY_PERCENTAGE` for percentage-based allocation.	`0`
`MEMORY_PERCENTAGE`	Percentage of total system memory to allocate. Used only when `MEMORY_CONSUMPTION=0`.	`50`
`TOTAL_CHAOS_DURATION`	Duration of the fault in seconds.	`60`
`INSTALL_DEPENDENCIES`	Install the stress tooling on the Windows host if missing.	`true`
`SEQUENCE`	Order in which multiple instances are stressed: `parallel` stresses all selected instances at once; `serial` stresses one at a time.	`parallel`
`RAMP_TIME`	Wait period in seconds before and after the fault. Go to ramp time to read how it is applied.	`0`

Authentication

Tunable	Description	Default
`ASSUME_ROLE_ARN`	ARN of an IAM role to assume on top of the base credentials. Leave empty to use the base credentials directly.	`""`
`AWS_AUTHENTICATION_SECRET`	Identifier of the File Secret in Harness Secret Manager that contains the AWS credentials file. Not required when using IRSA.	`""`

Tunables that apply to every fault are documented in common tunables for all faults. AWS-specific shared tunables are documented in common AWS fault tunables.

Fault execution in brief

Resolves the target instance list from EC2_INSTANCE_ID (or by EC2_INSTANCE_TAG), dispatches a PowerShell memory-allocation command via AWS Systems Manager Run Command to each selected Windows instance, holds the allocation for TOTAL_CHAOS_DURATION seconds, then releases the memory.

Expected behavior during fault execution

Free memory on the targeted Windows instances drops by approximately MEMORY_CONSUMPTION MB (or to the MEMORY_PERCENTAGE target).
If the working set exceeds available physical memory, Windows pages to the paging file; throughput collapses.
IIS and Windows services may experience increased response latency.
CloudWatch MemoryUtilization (when the CloudWatch Agent is installed) rises.

When the fault ends

The chaos pod releases the memory on each host. Memory returns to baseline within seconds.

Signals to watch

Attach resilience probes to assert each layer:

Memory utilization: Use a Prometheus probe on a memory-utilization metric (Memory % Committed Bytes In Use exported via CloudWatch).
Application latency: Use an HTTP probe and assert percentile latency SLOs.
Paging activity: Use a Prometheus probe on Pages/sec or equivalent to detect pressure.
Service availability: Monitor that Windows services stay in the Running state.

Verify the fault execution effect

While the experiment is running, confirm memory pressure is applied:

Inspect from the Windows host (via SSM Session Manager).
```
Get-Counter '\Memory\Available MBytes', '\Memory\% Committed Bytes In Use' -SampleInterval 1 -MaxSamples 5
```
Available memory should drop by approximately MEMORY_CONSUMPTION during the chaos window.
Check CloudWatch (if the CW Agent is installed).

In CloudWatch, mem_used_percent (or the equivalent) should rise on the affected instances.

Inspect SSM command status.

aws ssm list-command-invocations --region <region> --details --filters "key=Status,value=InProgress"

Recovery and cleanup

End of duration: The chaos pod releases the memory on each host.
Abort the experiment: Stopping the experiment from Chaos Studio cancels the SSM command.
Manual recovery: If the fault exits before cleanup, kill the stress process on the host via SSM Run Command (PowerShell Stop-Process).
Workload recovery: Pages swapped out during the stress are read back into memory on demand; some latency may persist briefly after recovery.

Limitations

Windows-only.
SSM-managed hosts only.
Paging file behaviour: If the paging file is small or disabled, Windows may BSOD or kill processes when memory is exhausted.
Cross-region targeting: A single experiment targets one region (the value of REGION).

Troubleshooting

Windows EC2 memory hog fails with AccessDeniedException in Harness Chaos Engineering

The credentials supplied to the chaos pod do not have the required EC2 or SSM permissions. Confirm the IAM policy attached to the user, role, or IRSA service account includes ec2:DescribeInstances, ssm:SendCommand, ssm:GetCommandInvocation, and the ec2messages:* actions used by SSM.

Windows EC2 memory hog reports instance is not SSM-managed

Confirm the SSM Agent is installed and running on the Windows host (the SSMAgent service should be in Running state), the host has an instance profile with AmazonSSMManagedInstanceCore (or equivalent), and the host appears in 'aws ssm describe-instance-information' with PlatformType=Windows.

System becomes unresponsive during the chaos window

If MEMORY_PERCENTAGE is very high (90+) and the paging file is small, Windows may freeze or BSOD as it runs out of memory. Lower MEMORY_PERCENTAGE to a value that leaves headroom (60-70 is usually safe), and ensure the paging file is sized appropriately for the workload.

Memory utilization does not rise during the chaos window

The most common causes are: the chaos process exited early (check the SSM command output); MEMORY_CONSUMPTION is small relative to total memory (raise it); or the host has more physical memory than expected. Check Get-Counter '\Memory\Available MBytes' on the host to confirm.

Windows EC2 CPU hog: Stress CPU on Windows instances instead of memory.
EC2 memory hog: Memory stress on Linux EC2 instances.
Windows EC2 process kill: Kill specific Windows processes on the instance.
Common AWS fault tunables: Shared environment variables for AWS faults.

Use cases​

Prerequisites​

Supported environments​

Permissions required​

Authentication​

Fault tunables​

Fault execution in brief​

Expected behavior during fault execution​

Signals to watch​

Verify the fault execution effect​

Recovery and cleanup​

Limitations​

Troubleshooting​

Related faults​