Skip to main content

A continuous test runner for gathering SLA data

Project description

SLA Monitor Worker

This is the test runner portion of the SLA monitor/reporter. It performs tests (or any command you want) repeatedly, and publishes success/failure to an SNS topic for external processing (for example, using lambda to write to a custom cloudwatch metric), as well as optionally uploading logs to an S3 bucket.

TODO: Unit tests not working


To install simply install via pip

pip install --user sla-runner

Highly recommended is iam-docker run:

pip install --user iam-docker-run

This project assumes you are using role based authentication, as would be used in a production environment in AWS. This mimics that environment by running with an actual role.


Excute the following in the root folder to run terraform. Obviously, have Terraform installed. Set the bucket and table variables to existing backend resources for remote state.

# pip install iam-starter
cd terraform
export AWS_ENV="dev"
export TF_DATA_DIR="./.$AWS_ENV-terraform/"
export AWS_DEFAULT_REGION="us-east-1"
export TF_STATE_REGION="us-east-1"
export TF_STATE_BUCKET="mycompany-tfstate-$AWS_ENV"
export TF_STATE_TABLE="tfstate_$AWS_ENV"

iam-starter \
    --profile $AWS_ENV \
    --command \
        "terraform init \
        -backend-config=\"region=$TF_STATE_REGION\" \
        -backend-config=\"bucket=$TF_STATE_BUCKET\" \
        -backend-config=\"dynamodb_table=$TF_STATE_TABLE\" && \
        terraform apply \
        -var \"aws_env=$AWS_ENV\" \
        -var \"aws_region=$AWS_DEFAULT_REGION\""


Use iam-docker-run outside of AWS to run tests. In real life scenarios on ECS, instead install sla-runner via pypi in your service container, and set --image to the image of the service container which contains your test.

docker build -t sla-monitor/sla-runner:latest .

export AWS_ENV="dev"
iam-docker-run \
    -e SLARUNNER_COMMAND="/bin/bash /src/test-scripts/" \
    -e SLARUNNER_SERVICE=example-service \
    -e SLARUNNER_GROUPS="dev-team,critical" \
    -e SLARUNNER_SNSTOPICNAME="sla-monitor-result-published-$AWS_ENV" \
    -e SLARUNNER_S3BUCKETNAME="sla-monitoring-logs-$AWS_ENV" \
    --full-entrypoint "sla-runner" \
    --region us-east-1 \
    --profile $AWS_ENV \
    --role sla-monitor-runner-role-$AWS_ENV \
    --image sla-monitor/sla-runner:latest

In ECS, add these as environment variables in the task definition or load from ssm via ssm-starter:

--full-entrypoint "ssm-starter --ssm-name slarunner --command 'sla-runner'"


The runner takes the following values which are provided by environment variable.

Global variables

When loading variables via SSM and ssm-starter, you can define default variables by adding a globals path before the service path.

For example, in your task definition in terraform:

    "entryPoint": ["ssm-starter"],
    "command": [
        "--ssm-name", "sla-monitor-globals",
        "--ssm-name", "${var.application}",
        "--command", "sla-runner" // or script that runs sla-runner



Command to be run repeatedly. Pretty straightforward. If there is an interrupt, the runner will attempt to finish the command gracefully before exit.



Name of the component service you're testing. This will be used as the prefix for s3 uploads, and will be passed in the JSON event as "service" to SNS.



Name of the grouping of components you're testing, in csv format. This will be passed in the JSON event as "groups" to SNS as a list, and is meant to provide secondary statistics if multiple services are part of the same component.



How long to wait between commands being run in seconds.



To disable sla-runner at startup.



SNS topic arn to publish results to. It will be published as a JSON object. For example, the command above would produce the following:

    "service": "example-service",
    "group": ["dev-team", "critical"],
    "succeeded": true,
    "timestamp": "1574515200",
    "testExecutionSecs": "914"



Bucket to write logs to. This is an optional parameter. The object will be named as the timestamp followed by the result for easily searching by result, and will be prefixed by the service name. For example "example-service/1574514000_SUCCESS"



If there is any value at all in this variable, the test will run once, output the sns topic it would publish to, the result message, the log output of the command, and the name of the object that would be written to the bucket. It will NOT publish to sns or write the object to s3. Only for testing purposes.

Development and Testing

docker build -t sla-runner:latest .
iam-docker-run \
    --image sla-runner:latest \
    --role sla-monitor-runner-role \
    --profile dev \
    --region us-east-1 \
    --host-source-path . \
    --container-source-path /src \

Publishing Updates to PyPi

For the maintainer - to publish an updated version of ssm-search, increment the version number in and run the following:

docker build -t sla-runner . &&
docker run --rm -it --entrypoint make sla-runner publish

At the prompts, enter the username and password to the repo.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

sla-runner-0.0.12.tar.gz (6.3 kB view hashes)

Uploaded source

Built Distribution

sla_runner-0.0.12-py3-none-any.whl (7.4 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page