ngx_template_module

ngx_template_module is the module for generating nginx runtime configuration with templates and inventory YAML files.

This module may be comparable with jinja2 templating engine with limited functionality.

On starting nginx ngx_template_module 'on the fly' generate configuration.

Build status

Requirements

ngx_template_module requires YAML library.

Status

Beta version.

Build

You may use build.sh script as a base. Simple run build.sh build and looking for a distributive in the install folder.

Directives

template

Syntax	template
Default	-
Syntax	template keyfile=inventory.yml [template=config.template]
Context	main,http,stream,server,location

Reads the inventory.yml file and transforms config.template into valid nginx configuration fragment.
template parameter may be absent. In this case inventory.yml file parsed and may be used in another templates.

All changes of templates and inventories are checked in background. When configuration is changed, ngx_template_module send reload signal to master nginx process.

Quick Start

This example uses directives from ngx_dynamic_healthcheck module.

Nginx main config fragment

template keyfile=env.yml template=env.template;

http {

    template keyfile=check.yml;
    template keyfile=servers.yml;

    template keyfile=backends.yml template=backend.template;
    template keyfile=entrypoints.yml template=entrypoint.template;

    server {
        listen 8888;

        location = /healthcheck/get {
            healthcheck_get;
        }

        location = /healthcheck/status {
            healthcheck_status;
        }
    }
}

Templates

Env template

env {{name}};

Upstream template

upstream {{ name }} {
    zone shm_{{ name }} 256k;

{{ servers@{{name}}.servers }}

    check passive type=http rise={{ check.rise | default({{ check@http.rise }}) }} fall={{ check.fall | default({{ check@http.fall }}) }} timeout={{ check.timeout | default({{ check@http.timeout }})}} interval={{ check.interval | default({{ check@http.interval }}) }};
    check_request_uri {{ check.request.method | default({{ check@http.request.method }}) }} {{ check.request.uri | default({{ check@http.request.uri }}) }};
    check_response_codes {{ check.response.codes | default({{ check@http.response.codes }}) }};
    check_response_body {{ check.response.match | default({{ check@http.response.match }}) }};
}

Server template

map 0 ${{name}} {
  {{map|default()}}
  default 0;
}

server {
    listen {{listen}} reuseport;

    default_type text/plain;

    location / {
        auth_request {{auth.request}};

        proxy_connect_timeout {{proxy.connect_timeout}};
        proxy_read_timeout {{proxy.read_timeout}};

        proxy_pass {{proxy.protocol}}://{{proxy.upstream|default({{name}})}};
    }

    location = /auth_stub {
        internal;
        return 200;
    }
}

Inventories

---
env:

  - name: LD_LIBRARY_PATH

---
check:

  - name: http
    rise: 1
    fall: 2
    timeout: 10000
    interval: 10
    request:
      method: GET
      uri: /health
    response:
      codes: 200 204
      match: .*

  - name: tcp
    rise: 1
    fall: 2
    timeout: 10000
    interval: 10

Iterating over all objects in array and applying template for each object.
All of parameters in YAML inventory MUST be a simple value or object (not an array). Arrays are unsupported except by in servers and map values.
Parameters may be defined as a macros. Ex:

entrypoints:

  - name: app1
    param: global_group@name.param
    listen: "{{env(APP1_LISTEN_PORT)|default(33333)}}"

where global_group is a top level element in YAML file. You may access to all of loaded inventories with group@name.xxx syntax usage. template directives must be declared in right order. This syntax may be used in templates and inventories.

Three upstreams will be created with directive template keyfile=backends.yml template=backend.template;.

---
backends:

  - name: app1
    server:
      max_conns: 10
      max_fails: 2
      fail_timeout: 30s
    check:
      rise: 1
      fall: 2
      timeout: 10000
      interval: 10
      request:
        method: GET
        uri: /health
      response:
        codes: 200 204
        match: .*

  - name: app2
    server:
      max_conns: 100
      max_fails: 2
      fail_timeout: 30s
    check:
      rise: 1
      fall: 2
      timeout: 10000
      interval: 10
      request:
        method: GET
        uri: /health
      response:
        codes: 200 204
        match: .*

  - name: app3
    server:
      max_conns: 100
      max_fails: 2
      fail_timeout: 30s

Three servers will be created with directive template keyfile=entrypoints.yml template=enptrypoint.template;.

---
entrypoints:

  - name: app1
    listen: "{{env(APP1_LISTEN_PORT)|default(33333)}}"
    auth:
      request: /auth_stub
    proxy:
      connect_timeout: 10s
      read_timeout: 60s
      protocol: http
      upstream: app1

  - name: app2
    listen: "{{env(APP2_LISTEN_PORT)|default(44444)}}"
    auth:
      request: /auth_stub
    proxy:
      connect_timeout: 10s
      read_timeout: 60s
      protocol: http

  - name: app3
    listen: 8081
    auth:
      request: /auth_stub
    proxy:
      connect_timeout: 10s
      read_timeout: 60s
      protocol: http
    map:
      - a 1;
      - b 2;
      - c 3;

Separate YAML file for all upstream's servers.

---
servers:

  - name: app1
    servers:
      - server 127.0.0.1:8001 max_conns={{backends@{{name}}.server.max_conns}};
      - server 127.0.0.1:8002 max_conns={{backends@{{name}}.server.max_conns}};

  - name: app2
    servers:
      - server 127.0.0.1:9001 max_conns={{backends@{{name}}.server.max_conns}};
      - server 127.0.0.1:9002 max_conns={{backends@{{name}}.server.max_conns}};

  - name: app3
    servers:
      - server 127.0.0.1:7001 max_conns={{backends@{{name}}.server.max_conns}};
      - server 127.0.0.1:7002 max_conns={{backends@{{name}}.server.max_conns}};

License

See LICENSE.

ZigzagAK / ngx_template_module