This project provides an infrastructure to automate the incident management of cloud-based services. With Amazon Web Services, we use state-of-the-art technologies to implement a variety of useful tools to alert, escalate, handle and resolve incidents. The core of our work is a chatbot which provides a human-like interface and natural language processing to enable the interaction to various web services concerning the incident management.
terraform plan
terraform apply
The alerting infrastructure needed for the devops-chatbot project can be automatically created using this terraform configuration. At the moment all the components from the diagram (architecture overview) plus the amazon lex-bot have been deployed and tested. Amazon Connect has been deployed and configured manually via the AWS console, since there is not support to automatically deploy/configure it.
Use -target to plan/apply single resources e.g.:
terraform plan -target=aws_dynamodb_table.escalation_target
Redeploy the kubernetes config:
terraform destroy -target=null_resource.Setup_Kubectl_Command
terraform apply -target=aws_lambda_function.Kubectl_Command
You can escalate or get the current incident via various ways.
- Text/Talk to the deployed 'DevOpsChatBot' lex-bot
- Call the phone number configured in AWS Connect
- Publish an appropriate message to the 'alert_from_cloudwatch' SNS topic(see the testcases folder for example message)
- Call the 'alert_manager_notification_api' API_GW with a POST request and appropriate request body/authentication(see the testcases folder)