neutron/doc/source/contributor/internals/services_and_agents.rst

Services and agents

A usual Neutron setup consists of multiple services and agents running on one or multiple nodes (though some exotic setups potentially may not need any agents). Each of those services provides some of the networking or API services. Among those of special interest:

1. neutron-server that provides API endpoints and serves as a single point of access to the database. It usually runs on nodes called Controllers.
2. Layer2 agent that can utilize Open vSwitch, Linuxbridge or other vendor specific technology to provide network segmentation and isolation for project networks. The L2 agent should run on every node where it is deemed responsible for wiring and securing virtual interfaces (usually both Compute and Network nodes).
3. Layer3 agent that runs on Network node and provides East-West and North-South routing plus some advanced services such as FWaaS or VPNaaS.

For the purpose of this document, we call all services, servers and agents that run on any node as just "services".

Entry points

Entry points for services are defined in setup.cfg under "console_scripts" section. Those entry points should generally point to main() functions located under neutron/cmd/... path.

Note: some existing vendor/plugin agents still maintain their entry points in other locations. Developers responsible for those agents are welcome to apply the guideline above.

Interacting with Eventlet

Neutron extensively utilizes the eventlet library to provide asynchronous concurrency model to its services. To utilize it correctly, the following should be kept in mind.

If a service utilizes the eventlet library, then it should not call eventlet.monkey_patch() directly but instead maintain its entry point main() function under neutron/cmd/eventlet/... If that is the case, the standard Python library will be automatically patched for the service on entry point import (monkey patching is done inside python package file).

Note: an entry point 'main()' function may just be an indirection to a real callable located elsewhere, as is done for reference services such as DHCP, L3 and the neutron-server.

For more info on the rationale behind the code tree setup, see the corresponding cross-project spec.

Connecting to the Database

Only the neutron-server connects to the neutron database. Agents may never connect directly to the database, as this would break the ability to do rolling upgrades.

Configuration Options

In addition to database access, configuration options are segregated between neutron-server and agents. Both services and agents may load the main `neutron.conf since this file should contain the oslo.messaging configuration for internal Neutron RPCs and may contain host specific configuration such as file paths. In addition neutron.conf` contains the database, Keystone, and Nova credentials and endpoints strictly for neutron-server to use.

In addition neutron-server may load a plugin specific configuration file, yet the agents should not. As the plugin configuration is primarily site wide options and the plugin provides the persistence layer for Neutron, agents should be instructed to act upon these values via RPC.

Each individual agent may have its own configuration file. This file should be loaded after the main `neutron.conf` file, so the agent configuration takes precedence. The agent specific configuration may contain configurations which vary between hosts in a Neutron deployment such as the external_network_bridge for a L3 agent. If any agent requires access to additional external services beyond the Neutron RPC, those endpoints should be defined in the agent specific configuration file (e.g. nova metadata for metadata agent).