Open folder
Open the project folder in the chat.
Workflows
dns load balancing and global load balancer design
03
Design a pool with health checks
You need a pool of backends that share traffic. The agent reads your notes, drafts the configuration file, and runs your test command to verify the syntax before you apply it.
Read config
Read the current config file to see existing backends.
Draft changes
Write the new pool configuration with health checks.
Verify syntax
Run the config test command to check for errors.
Example configuration
Draft a pool config with three backends and a health check.
- write_file
lb-plan.yamlsaved - list_dir
.1 file - read_file
lb-plan.yamlread 12 lines - grep
backend3 matches
waiting for your OK before the next stepRejectAccept
lb-plan.yaml
# Load balancer pool config
# Algorithm: round robin
pool:
name: web-pool
algorithm: round_robin
backends:
- addr: 10.0.0.1
port: 8080
- addr: 10.0.0.2
port: 8080
- addr: 10.0.0.3
port: 8080
health_check:
path: /health
interval: 10s
fail_threshold: 3
pass_threshold: 2
standby:
- addr: 10.0.0.4
port: 8080This yaml file defines a pool with three active backends and one standby server, including health check parameters.
Verify failover on your own systems
Draft the failover plan
The agent uses list_dir to see your folder structure and read_file to inspect your current notes or configuration files so you understand the layout. grep finds the specific ports. edit_file adds a standby server to handle traffic if one backend fails. You see a diff card showing exactly what will change in the pool. Shell runs your configuration test command and reports the exit code. Switch to Plan only mode to see the full sequence of changes without executing them. dns load balancing helps manage traffic, and you can verify dual wan failover with static ip during testing. Each action appears in the chat as a card with its real result.
Test the failover manually
The reader stops one backend in a planned window and the agent uses http_get to check the health page of the remaining backends. It checks the stopped backend. The agent uses write_file to update your runbook with the test results. DNS caching can delay failover for the TTL duration significantly. The agent records these results in the chat interface clearly. This verifies that the global load balancer configuration is correct. You verify traffic failover to another link by starting the backend again now. The agent checks that it returns to the pool fully.
