Helper script to verify state of HAProxy backend servers. Use this to verify the state of HAProxy backends after you drain or enable it in your pipeline.
HAProxy typically has multiple backends which contain pool of servers to load-balance requests.
To deploy to a particular server, sysadmins need to drain the server of incoming requests before deploy the changes (applicable if you are not on k8s yet). This manual step can be automated by enabling Unix socket commands or through rest api in recent HAProxy versions.
This repo relies on Unix socket commands being enabled, it can easily be refactored to work with REST API.
A typical workflow would look like the following:
You are deploying to ServerX
Step 1. Drain ServerX in haproxy via socat command/rest api
Step 2. Verify ServerX backend server is drained
Step 3. Tear down and/or deploy ServerX
Step 4. Enable ServiceX in haproxy via socat command/rest api
The check server state script is intented to live in the same place that has HAProxy and is called by CI/CD pipeline e.g. Jenkins to validate servers are drained/enabled.
The script can live elsewhere if REST API is used to communicate with HAPRoxy. The design depends on the nature of your project and deployment patterns. Socat works nicely if you use CHEF as chef-client runs directly on the target node.
check_server_state.py --backend ServerX --drain
will poll the current sessions for the server until it drops to 0 or timeout is reached. This should be used as a verification step in the CI/CD pipeline after setting the server to drain mode and prior to deploying.
check_server_state.py --backend ServerX --ready
will check the server is in UP mode. This should be used as a verification step in the CI/CD pipeline after deploying and enabling the server backend.
Run check_server_statey.py --help for info on arguments to provide.