main.zeek

Management::Controller::Runtime

This is the main “runtime” of the Management framework’s controller. Zeek does not load this directly; rather, the controller’s bootstrapping module (in ./boot.zeek) specifies it as the script to run in the node newly created by the supervisor.

Namespace:: Management::Controller::Runtime
Imports:: base/frameworks/broker, policy/frameworks/management, policy/frameworks/management/agent/api.zeek, policy/frameworks/management/agent/config.zeek, policy/frameworks/management/controller/api.zeek, policy/frameworks/management/controller/config.zeek

Summary

Types

`Management::Controller::Runtime::ConfigState`: `enum`	A cluster configuration uploaded by the client goes through multiple states on its way to deployment.
`Management::Controller::Runtime::DeployState`: `record`	Request state specific to `Management::Controller::API::deploy_request` and `Management::Controller::API::deploy_response`.
`Management::Controller::Runtime::GetNodesState`: `record`	Request state specific to `Management::Controller::API::get_nodes_request` and `Management::Controller::API::get_nodes_response`.
`Management::Controller::Runtime::NodeDispatchState`: `record`	Request state for node dispatch requests, to track the requested action and received responses.
`Management::Controller::Runtime::RestartState`: `record`	Request state specific to `Management::Controller::API::restart_request` and `Management::Controller::API::restart_response`.
`Management::Controller::Runtime::TestState`: `record`	Dummy state for internal state-keeping test cases.

Redefinitions

`Management::Request::Request`: `record`	New Fields: `Management::Request::Request` deploy_state: `Management::Controller::Runtime::DeployState` `&optional` get_nodes_state: `Management::Controller::Runtime::GetNodesState` `&optional` node_dispatch_state: `Management::Controller::Runtime::NodeDispatchState` `&optional` restart_state: `Management::Controller::Runtime::RestartState` `&optional` test_state: `Management::Controller::Runtime::TestState` `&optional`
`Management::role`: `Management::Role` `&redef`
`table_expire_interval`: `interval` `&redef`

Detailed Interface

Types

Management::Controller::Runtime::ConfigState 

Type:

enum

Management::Controller::Runtime::STAGED: As provided by the client.

Management::Controller::Runtime::READY: Necessary updates made, e.g. ports filled in.

Management::Controller::Runtime::DEPLOYED: Sent off to the agents for deployment.

A cluster configuration uploaded by the client goes through multiple states on its way to deployment.

Management::Controller::Runtime::DeployState 

Type:

record

config: Management::Configuration: The cluster configuration the controller is deploying.
is_internal: bool &default = F &optional: Whether this is a controller-internal deployment, or triggered via a request by a remote peer/client.
requests: set [string] &default = { } &optional: Request state for every controller/agent transaction.

Request state specific to Management::Controller::API::deploy_request and Management::Controller::API::deploy_response.

Management::Controller::Runtime::GetNodesState 

Type:

record

requests: set [string] &default = { } &optional: Request state for every controller/agent transaction.

Request state specific to Management::Controller::API::get_nodes_request and Management::Controller::API::get_nodes_response.

Management::Controller::Runtime::NodeDispatchState 

Type:

record

action: vector of string: The dispatched action. The first string is a command, any remaining strings its arguments.
requests: set [string] &default = { } &optional: Request state for every controller/agent transaction. The set of strings tracks the node names from which we still expect responses, before we can respond back to the client.

Request state for node dispatch requests, to track the requested action and received responses. Node dispatches are requests to execute pre-implemented actions on every node in the cluster, and report their outcomes. See Management::Agent::API::node_dispatch_request and Management::Agent::API::node_dispatch_response for the agent/controller interaction, and Management::Controller::API::get_id_value_request and Management::Controller::API::get_id_value_response for an example of a specific API the controller generalizes into a dispatch.

Management::Controller::Runtime::RestartState 

Type:

record

requests: set [string] &default = { } &optional: Request state for every controller/agent transaction.

Request state specific to Management::Controller::API::restart_request and Management::Controller::API::restart_response.

Management::Controller::Runtime::TestState 

Type:: record

Dummy state for internal state-keeping test cases.