Service Startup Dependencies ​

lzcapp provides two fields related to service dependencies to handle timing issues during application startup

1. Simplified depends_on field
2. Health check mechanism automatically injected based on application.routes field

depends_on ​

In most cases, depends_on only needs to be used under the services.$service_name field, type is []string Each entry fills in a $service_name, the current service will wait for all services in depends_on to have container status as healthy when starting

The service.$service_name.health_check field will affect the service entering healthy state. This field has the same semantics as docker-compose's healthCheck, but only supports the following 4 fields

type HealthCheckConfig struct {
	Test        []string      `yaml:"test"`
	TestUrl     string        `yaml:"test_url"`
	StartPeriod time.Duration `yaml:"start_period"`
	Disable     bool          `yaml:"disable"`
}

type HealthCheckConfig struct {
	Test        []string      `yaml:"test"`
	TestUrl     string        `yaml:"test_url"`
	StartPeriod time.Duration `yaml:"start_period"`
	Disable     bool          `yaml:"disable"`
}

health_check example:

services:
  wordpress:
    image: bitnami/wordpress:5.8.2
    health_check:
        test:
            - CMD-SHELL
            - "curl -f http://localhost:80/ || exit 1"
        # test_url: http://localhost:80 # URL for health check, health check fails if status code is greater than 500
        start_period: 190s
        disable: false

services:
  wordpress:
    image: bitnami/wordpress:5.8.2
    health_check:
        test:
            - CMD-SHELL
            - "curl -f http://localhost:80/ || exit 1"
        # test_url: http://localhost:80 # URL for health check, health check fails if status code is greater than 500
        start_period: 190s
        disable: false

Automatically Injected Health Checks ​

lzcapp mainly provides services externally through application.routes, so the system will automatically perform intelligent detection based on the upstream state in routes. Therefore, in most cases, there is no need to manually handle dependencies. The specific rules are as follows:

1. Detect and wait for all service containers to enter running state
2. Detect and wait for the corresponding service in application.health_check.test_url to return 200 status (if this configuration exists)
3. Detect and wait for all upstreams to be ready, then the special service application enters healthy state The specific method is to scan all routes rules and extract the http://$hostname:$port part. exec type will be automatically converted to http://127.0.0.1:$port Use TCP dial $hostname:$port, if dial succeeds, it indicates that this upstream is ready
4. Wait for all other services to enter healthy state, the "Application starting" page switches to actual service content