Lambda inject latency

Inject runtime latency into an AWS Lambda function for a configurable duration so you can test how upstream callers and downstream consumers handle slower-than-expected responses, cold-start spikes, a

Lambda inject latency is an AWS chaos fault that adds LAMBDA_LATENCY seconds of delay to every invocation of a target Lambda function (FUNCTION_NAME in REGION) for TOTAL_CHAOS_DURATION seconds. The fault injects a runtime wrapper that sleeps for the configured latency on each invocation; at the end of the window it restores the original function configuration.

Use this fault to test how an event-driven or API-driven workload behaves when a Lambda function becomes slow: whether upstream callers honor timeouts, whether retries amplify the load, whether monitoring detects the latency increase, and whether downstream consumers degrade gracefully under stretched response times.

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:

• Caller timeout behaviour: When the function takes longer than expected, do upstream callers (API Gateway, ALB, EventBridge, Step Functions, application code) honor their configured timeouts?

• Function timeout exposure: Does the function's own timeout protect against slow downstreams, or does it amplify cost and consumer latency?

• Retry storms: Do callers retry latent invocations and amplify load on the function?

• Monitoring fidelity: Do CloudWatch alarms on Duration, ConcurrentExecutions, and downstream p99 fire within the SLA?

• Capacity headroom: Does extended duration push the function past its reserved concurrency and trigger throttling?

---

Prerequisites

• Kubernetes version: 1.21 or later for the chaos infrastructure cluster. Go to What's supported to confirm distribution support.

• Target Lambda function: FUNCTION_NAME exists in REGION and is in the Active state.

• Latency vs function timeout: Confirm LAMBDA_LATENCY is smaller than the function's configured timeout, otherwise every invocation hits the function timeout instead of completing.

• 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
AWS Lambda (zip and container image)	Supported
Lambda with provisioned concurrency	Supported
Lambda@Edge	Not supported
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": [
        "lambda:GetFunction",
        "lambda:GetFunctionConfiguration",
        "lambda:UpdateFunctionConfiguration",
        "lambda:UpdateFunctionCode",
        "lambda:PublishVersion"
      ],
      "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.

---

Fault tunables

Configure the following fault parameters when you add Lambda inject latency to an experiment in Chaos Studio. Defaults are shown for reference.

Required parameters

Tunable	Description	Default
FUNCTION_NAME	Name of the target Lambda function.	(required)
REGION	AWS region where the Lambda function is deployed.	(required)
LAMBDA_LATENCY	Latency to inject into every invocation, in seconds. Keep this lower than the function's configured timeout.	5

Chaos parameters

Tunable	Description	Default
TOTAL_CHAOS_DURATION	Duration of the fault in seconds.	30
CHAOS_INTERVAL	Delay in seconds between successive iterations when running for more than one cycle.	30
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.

---

Fault execution in brief

Reads the current function configuration, injects a runtime wrapper that sleeps for LAMBDA_LATENCY seconds on each invocation, waits for TOTAL_CHAOS_DURATION seconds, then restores the original function configuration.

---

Expected behavior during fault execution

• Every new invocation of FUNCTION_NAME takes at least LAMBDA_LATENCY seconds longer to complete.

• CloudWatch metric Duration for the function rises by approximately LAMBDA_LATENCY * 1000 ms.

• If the function is synchronously invoked (API Gateway, ALB, application code), callers see slower responses or timeouts.

• If LAMBDA_LATENCY is close to the function timeout, Errors and Throttles may rise as invocations exceed the configured timeout or push concurrency past its limit.

WHEN THE FAULT ENDS

The chaos pod restores the original function configuration. New invocations return to baseline latency. Warm containers that carry the wrapper are recycled on the next cold start.

Signals to watch

Attach resilience probes to assert each layer:

• Lambda duration: Use a Prometheus probe on the duration metric for the function.

• Lambda errors/throttles: Use Prometheus probes on aws_lambda_errors_sum and aws_lambda_throttles_sum.

• End-to-end latency: Use an HTTP probe on the user-visible endpoint and assert p95 / p99 stays inside the SLO.

• Caller behaviour: Use an HTTP probe with the caller's timeout to confirm timeouts are honored.

---

Verify the fault execution effect

While the experiment is running, confirm the function is slower and then recovers:

1. Invoke the function and measure round-trip time.

   time aws lambda invoke --region <region> --function-name <name> /tmp/out.json

   Round-trip should rise by approximately LAMBDA_LATENCY seconds during the chaos window and return to baseline afterwards.

2. Check CloudWatch Duration.

   aws cloudwatch get-metric-statistics --region <region> \
     --namespace AWS/Lambda --metric-name Duration \
     --dimensions Name=FunctionName,Value=<name> \
     --start-time <iso> --end-time <iso> \
     --period 60 --statistics Average,Maximum

   Average and maximum should rise by ~LAMBDA_LATENCY * 1000 ms during the chaos window.

3. Inspect logs.

   aws logs tail --region <region> /aws/lambda/<name> --since 5m

   REPORT lines should show longer billed durations during the chaos window.

---

Recovery and cleanup

• End of duration: The chaos pod restores the original function configuration.

• Abort the experiment: Stopping the experiment from Chaos Studio also triggers the restore call.

• Manual recovery: If the fault exits before restore, run aws lambda update-function-configuration --function-name <name> --region <region> with the original environment/runtime values (recorded in the chaos pod logs).

• Workload recovery: Warm containers carrying the wrapper are replaced on the next cold start; until then, sporadic latency may continue on warm invocations of the same execution environment.

---

Limitations

• Per-invocation only: The fault adds latency on every invocation; it cannot target a percentage of invocations or specific code paths.

• Cold start interaction: Updating the function configuration triggers cold starts during the chaos window; the first invocations after the start may include both LAMBDA_LATENCY and the cold-start init time.

• Latency vs timeout: If LAMBDA_LATENCY exceeds the function timeout, every invocation fails rather than completing slowly.

• Cross-region targeting: A single experiment targets one region (the value of REGION).

---

Troubleshooting

Lambda inject latency fails with AccessDeniedException in Harness Chaos Engineering

The credentials supplied to the chaos pod do not have the required Lambda permissions. Confirm the IAM policy attached to the user, role, or IRSA service account includes lambda:GetFunction, lambda:GetFunctionConfiguration, lambda:UpdateFunctionConfiguration, lambda:UpdateFunctionCode, and lambda:PublishVersion.

Function invocations time out instead of returning slowly

LAMBDA_LATENCY is greater than or close to the function's configured timeout. Reduce LAMBDA_LATENCY below the function timeout, or temporarily raise the function timeout before the experiment to give invocations enough headroom to complete.

No latency change observed in CloudWatch metrics

If the function is invoked rarely or asynchronously and metrics are averaged over a long period, the added latency may be diluted. Drive sustained synchronous load against the function during the chaos window and re-check the Duration metric on a 1-minute period.

Function configuration not restored after the chaos window

If the restore call failed, the function may still carry the wrapper. Use 'aws lambda get-function-configuration --function-name --region ' to inspect the current configuration and roll it back to the original values recorded in the chaos pod logs.

---

Related faults

• Lambda inject status code: Override HTTP status codes instead of adding latency.

• Lambda modify response body: Override response bodies.

• Lambda block TCP connection: Block outbound TCP connections from the function.

• Common AWS fault tunables: Shared environment variables for AWS faults.