Wanda

A service responsible to orchestrate Checks executions on a target infrastructure.

Documentation

The documentation is available at trento-project.io/wanda.

Swagger UI is available at trento-project.io/wanda/swaggerui.

Developing Checks

Wanda architecture aims to simplify testing Checks Executions and adding new ones.

Infrastructure

For development purposes, a docker-compose file is provided. The docker-compose.checks.yaml provides additional configuration to start an environment for Checks development.

Starting a local environment

Start the environment with:

$ docker-compose -f docker-compose.checks.yaml up -d

Wanda is exposed on port 4000 and the API documentation is available at http://localhost:4000/swaggerui

Note that the message broker must be reachable by Wanda and all the targets.

Testing Executions

With a running setup, it is possible to easily test Checks and their Execution by:

starting the targets
consulting the catalog
starting a Checks Execution
checking the state of an execution
debugging the gathered facts

Starting the targets

The trento-agent must be up and running on the targets to run a correct execution, otherwise a timeout error is raised.

Here an example on how to start it:

./trento-agent start --api-key <api-key> --facts-service-url amqp://wanda:wanda@localhost:5674

Note: api-key value is not used if the unique goal is to run checks, so setting it as --api-key 0 does the work.

Keep in mind that the agent_id of the targets must match with values provided in the targets field of the execution request.

The ID can be obtained running:

./trento-agent id

If the execution is run in a development/testing environment, faking the agent id might come handy.

Consulting the catalog

Available Checks are part of the Catalog, and they can be retrieved by accessing the dedicated API

curl -X 'GET' \
  'http://localhost:4000/api/v1/checks/catalog' \
  -H 'accept: application/json'

Starting a Checks Execution

A Checks Execution can be started by calling the Start Execution endpoint, as follows

curl --request POST 'http://localhost:4000/api/v1/checks/executions/start' \
--header 'accept: application/json' \
--header 'Content-Type: application/json' \
--data-raw '{
  "env": {
    "provider": "azure"
  },
  "execution_id": "205e326d-0c25-4f4b-9976-43f9ba1c86d3",
  "group_id": "3dff9d03-4adf-453e-9513-8533e221bb12",
  "targets": [
    {
      "agent_id": "a644919a-d953-43d4-bd57-7e0bb96ee894",
      "checks": [
        "156F64"
      ]
    },
    {
      "agent_id": "02d99b2f-0efd-443c-ac9c-32710323f620",
      "checks": [
        "OTH3R1"
      ]
    }
  ]
}'

execution_id must be new and unique for every new execution. If an already used execution_id is provided, starting the execution fails.

In order to get detailed information for an execution, see Getting Execution details.

Note that execution is eventually started, meaning that a successful response to the previous API call does not guarantee that the execution is running, but that it has been accepted by the system to start.

Execution Targets

An execution target is a target host where the checks are executed. This requires to have the trento-agent executable running in the host. In order to specify an execution order, its agent_id and a list of checks to be executed are provided. Once the execution is started, a facts gathering request is sent to these targets, facts are gathered and sent back to Wanda, where the checks result is evaluated using the gathered facts.

The agent_id can be obtained just issuing trento-agent id command in the target host.

Each target must specify a list of checks, that can be empty. These are the selected checks for each agent, that are executed.

Given two different targets, the same checks can be selected:

curl --request POST 'http://localhost:4000/api/v1/checks/executions/start' \
--header 'accept: application/json' \
--header 'Content-Type: application/json' \
--data-raw '{
  "env": {
    "provider": "azure"
  },
  "execution_id": "205e326d-0c25-4f4b-9976-43f9ba1c86d3",
  "group_id": "3dff9d03-4adf-453e-9513-8533e221bb12",
  "targets": [
    {
      "agent_id": "a644919a-d953-43d4-bd57-7e0bb96ee894",
      "checks": [
        "156F64",
        "45B653"
      ]
    },
    {
      "agent_id": "02d99b2f-0efd-443c-ac9c-32710323f620",
      "checks": [
        "156F64",
        "45B653"
      ]
    }
  ]
}'

Or completely different ones:

curl --request POST 'http://localhost:4000/api/v1/checks/executions/start' \
--header 'accept: application/json' \
--header 'Content-Type: application/json' \
--data-raw '{
  "env": {
    "provider": "azure"
  },
  "execution_id": "205e326d-0c25-4f4b-9976-43f9ba1c86d3",
  "group_id": "3dff9d03-4adf-453e-9513-8533e221bb12",
  "targets": [
    {
      "agent_id": "a644919a-d953-43d4-bd57-7e0bb96ee894",
      "checks": [
        "156F64",
        "45B653"
      ]
    },
    {
      "agent_id": "02d99b2f-0efd-443c-ac9c-32710323f620",
      "checks": [
        "OTH3R1",
        "OTH3R2"
      ]
    }
  ]
}'

Getting Execution details

To get detailed information about the execution, the following API can be used.

curl --request GET 'http://localhost:4000/api/v1/checks/executions/205e326d-0c25-4f4b-9976-43f9ba1c86d3' \
--header 'accept: application/json' \
--header 'Content-Type: application/json'

Note that calling the execution detail API right after starting an execution might result in a 404 not found, because the execution, as mentioned, is eventually started.

In this case retry getting the detail of the execution.

Refer to the API doc for more information about requests and responses.

Debugging gathered facts

Often times knowing the returned value of the gathered facts is not a trivial thing, more during the implementation of new checks.

To better debug the fact gathering process and the returned values, the facts subcommand of trento-agent is a really useful tool. This command helps to see in the target itself what the gathered fact looks like. This is specially interesting when the returned value is a complex object or the target under test is modified and the check developer wants to see how this affects the gathered fact.

The command can be used as:

./trento-agent facts gather --gatherer corosync.conf --argument totem.token
# To see the currently available gatherers and their names
# ./trento-agent facts list

Which would return the next where the Value is the available value in the written check:

{
  "Name": "totem.token",
  "CheckID": "",
  "Value": {
    "Value": 30000
  },
  "Error": null
}

Adding new Checks

Built-in Checks can be found in the Catalog directory at ./priv/catalog/

To implement new checks and test them:

write a new Check Specification file
locate the newly created Check in the Catalog directory ./priv/catalog/
test the execution as previously described

Installing a development environment

To set up a local development environment for hacking on Wanda, you need to follow the instructions provided in hack on wanda.

This guide walks you through the process of installing and configuring the necessary dependencies, as well as setting up a local development environment.

Support

Please only report bugs via GitHub issues and for any other inquiry or topic use GitHub discussion.

Contributing

See CONTRIBUTING.md

License

Licensed under the Apache License, Version 2.0 (the "License"); you may not use any of the source code in this project except in compliance with the License. You may obtain a copy of the License at

https://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

yeoldegrove/wanda