27 releases

0.4.0-pre1 Oct 3, 2022
0.3.2 Jul 10, 2022
0.3.0-alpha.8 Feb 7, 2022
0.2.5 Dec 23, 2021
0.2.3-alpha.5 Nov 30, 2021

#899 in Web programming

Download history 7/week @ 2024-09-20 2/week @ 2024-09-27

74 downloads per month

Apache-2.0

170KB
3K SLoC

Rust 2.5K SLoC // 0.1% comments Python 358 SLoC // 0.4% comments Jinja2 86 SLoC SQL 2 SLoC // 0.8% comments

About

mx-tester is a WIP tool to help test Matrix bots and Synapse modules.

What mx-tester does

The flow of mx-tester is the following:

  1. mx-tester build Create a Docker image for Synapse. If you need your Synapse to be setup with custom modules, you add scripts to this step to install the modules.
  2. mx-tester up Launch Synapse on port 9999.
  3. mx-tester run Launch your test script.
  4. mx-tester down Stop Synapse.

You may customize each of these steps by inserting scripts.

If you are using mx-tester to launch tests on a project that has already been setup, your flow is probably simply:

# Setup and launch Synapse, modules, bots, ...
$ mx-tester build
$ mx-tester up
# Launch the tests. Repeat as many times as needed, as you fix tests or your code.
$ mx-tester run
# Once you're done testing
$ mx-tester down

or in a single go

$ mx-tester build up run down

Setting up mx-tester.

mx-tester requires a configuration file, typically called mx-tester.yml.

It has the following structure:

name: A name for this test suite

# --- Configuring external components

up:
  # Optional. A script to be executed to prepare additional components
  # as part of `mx-tester up`.
  before:
    - # Optional. Script to be executed before bringinng up the image.
    - # Use this script e.g. to setup databases, etc.
  after:
    - # Optional. Script to be executed after bringing up the image.
    - # Use this script e.g. to setup bots.

down:
  # Optional. A script to be executed to clean up the environment,
  # typically to teardown components that were setup with `up`. This
  # script will be executed at the end of `mx-tester down`.
  success:
    - # Optional. A script to be executed only if `run` was a success.
    - # This script will ONLY be executed if `mx-tester` is called to
    - # execute both run and down, e.g. `mx-tester up run down`.
  failure:
    - # Optional. A script to be executed only if `run` was a failure.
    - # This script will ONLY be executed if `mx-tester` is called to
    - # execute both run and down, e.g. `mx-tester up run down`.
  finally:
    - # Optional. A script to be executed regardless of the result of `run`.
    - # This script will ALWAYS be executed if `mx-tester down` is called.

directories:
  # Optional. Directories to use for the test.
  root:
    # Optional. The root directory for this test.
    # All temporary files and logs are created as subdirectories of this directory.
    # Default: `mx-tester` in the platform's temporary directory.
    # May be overridden from the command-line with parameter `--root`.
    #
    # IMPORTANT: Some CI environments (e.g. Docker-in-docker aka dind) do not play
    # well with the platform's temporary directory. If you are running `mx-tester`
    # in such an environment, you should probably specify your root directory to
    # be in your (environment-specific) home-style directory (which may be your
    # build directory, etc.).
    #
    # Example: On GitLab CI, you should probably use `/builds/$CI_PROJECT_PATH`.

# --- Configuring the test

run:
  - # Optional. A script to be executed as `mx-tester run`.
  - # Use this e.g. to launch integration tests.

users:
  - # Optional. A list of users to create for the test.
  - # These users are created during `mx-tester up`.
  - # If the users already exist, they are not recreated.
  - localname:
    # Required. A name for the user.
    admin:
    # Optional. If `true`, the users should be an admin for the server.
    # Default: `false`.
    password:
    # Optional. If specified, a password for the user.
    # Default: "password".
    rate_limit:
    # Optional. If `unlimited`, remove rate limits for this user.
    # Default: Use the global setting for rate limits.
    rooms:
    - # Optional. A list of rooms to create.
    - public:
      # Optional. If `true`, the room should be public.
      # Default: `false`.
      name:
      # Optional. A name for the room.
      # Default: No name.
      alias:
      # Optional. An alias for the room.
      # Default: No alias.
      # If there is already a room with the same alias, the old alias will
      # be unregistered (we assume that this was caused by a previous call
      # to `mx-tester up`).
      # It several rooms define the same `alias` in the same mx-tester.yml,
      # this is an error.
      topic:
      # Optional. A topic for the room.
      # Default: No topic.
      members:
      # Optional. A list of users (created by `users`) to invite to the room.
      # mx-tester will ensure that these users join the room.
      # Default: No invites.


# --- Configuring the homeserver. Optional.

synapse:
  # Optionally, a version of Synapse.
  # If unspecified, pick the latest version available on Docker Hub.
  docker:
    # Required: A docker tag, e.g. "matrixdotorg/synapse:latest"

modules:
  # Optionally, a list of modules to install.
  - name: Name of a module you wish to setup
    build:
      # Required: A script to setup the module.
      # This may be as simple as copying the module from its directory
      # to $MX_TEST_MODULE_DIR.
      - # This script MUST copy the source code of the module
      - # to directory $MX_TEST_MODULE_DIR
      - # ...
      - # env: MX_TEST_MODULE_DIR -- where the module should be copied
      - # env: MX_TEST_SYNAPSE_DIR -- where Synapse source lies
      - # env: MX_TEST_SCRIPT_TMPDIR -- a temporary directory where the test can
      - #   write data. Note that `mx-tester` will NOT clear this directory.
      - # env: MX_TEST_CWD -- the directory in which the test was launched.
    install:
      # Optional. A script to install dependencies.
      # Typically, this will be something along the lines of
      # `pypi -r module_name/requirements.txt`
    config:
      # Required. Additional configuration information
      # to copy into homeserver.yaml.
      # This typically looks like
      # module: python_module_name
      # config:
      #   key: value
      #   key: value
      #   ...
  - # Other modules, if necessary.

homeserver:
  # Optional. Additional configuration for the homeserver.
  # Each of these fields will be copied into homeserver.yaml.
  # For more detail on the fields, see the documentation for homeserver.yaml.
  server_name:
    # Optional. The name of a homeserver.
    # By default, `localhost:9999`.
  public_baseurl:
    # Optional. The URL to communicate to the server with.
    # By default, `http://localhost:9999`.
  registration_shared_secret:
    # Optional. The registration shared secret.
    # By default, "MX_TESTER_REGISTRATION_DEFAULT".
  ...
    # Any other field to be copied in homeserver.yaml.

appservices:
  # Optional. Define Matrix Application Services to be connected
  # with your homeserver. For more details on AppServices, see
  # https://spec.matrix.org/v1.2/application-service-api .
  host:
    # Optional. Application Services whose process is running on
    # the host.
    name:
      # String. The name of the AppService. Each AppService on this
      # homeserver should have a distinct name.
    url:
      # String. The (local) URL at which to communicate with the AppService,
      # including a port number, e.g. "http://localhost:8765".
      # Note that the host is actually IGNORED.
    as_token:
      # Optional. String. If specified, a token used by the homeserver
      # when sending messages to the AppService.
      # By default, generate a random token automatically during `build`.
      # See the generated appservice file to determine its value.
    hs_token:
      # Optional. String. If specified, a token used by the AppService
      # when sending messages to the homeserver.
      # By default, generate a random token automatically during `build`.
      # See the generated appservice file to determine its value.
    sender_localpart:
      # String. The localpart of the userid of the user that will be
      # used when this AppService sends messages.
    namespaces:
      # The namespaces controlled or monitored by this AppService.
      users:
        # Optional. Empty by default.
        - regex:
            # A set of users to control or monitor.
            # Regex are matched against the full user id, including sigil
            # and homeserver name.
          exclusive:
            # Optional. Boolean. False by default.
            # If `true`, the AppService is in charge of all users whose
            # id matches the regex. Otherwise, the AppService will only
            # be informed of the actions of these users.
      rooms:
        # Optional. Empty by default.
        - regex:
            # A set of rooms to control or monitor.
            # Regex are matched against the full room id, including sigil
            # and homeserver name.
          exclusive:
            # Optional. Boolean. False by default.
            # If `true`, the AppService is in charge of all rooms whose
            # id matches the regex. Otherwise, the AppService will only
            # be informed of the actions in these rooms.
      aliases:
        # Optional. Empty by default.
        - regex:
            # A set of room alises to control or monitor.
            # Regex are matched against the full alias, including sigil
            # and homeserver name.
          exclusive:
            # Optional. Boolean. False by default.
            # If `true`, the AppService is in charge of all rooms whose
            # alias matches the regex. Otherwise, the AppService will only
            # be informed of the actions in these aliases.
  guest:
    # Optional. Application Services whose process is running on the guest.
    # Same fields as `host`.

# --- Additional resources. Optional.

copy:
  # Optional. Additional resources to copy.
  # Paths on the guest are relative to the guest root.
  # Paths on the host are relative to the directory from which mx-tester
  # is launched.
  # As of this writing, only individual files are copied, no globs or
  # subdirectories.
  # Guest directories are created if necessary.
  "/path/on/the/guest": "path/on/the/host"
  # ...

# --- Docker configuration. Optional.

docker:
  # Optional. Additional configuration for Docker.
  hostname:
    # Optional. The hostname to give to the Synapse container on the docker
    # network.
    # By default, "synapse".
  port_mapping:
    - # Optional. The docker port mapping configuration to use for the
    - # synapse container.
    - # By default:
    - # - host: 9999
    - # - guest: 8008

credentials:
  # Optional. Credentials to connect to a Docker registry,
  # as per `docker login`.
  # Typically useful to adapt mx-tester to your CI.
  username:
  # Optional. Specify a username to connect to the registry.
  # Default: No username.
  # May be overridden from the command-line with parameter `--username`/`-u`.
  password:
  # Optional. Specify a password to connect to the registry.
  # Default: No password.
  # May be overridden from the command-line with parameter `--password`/`-p`.
  serveraddress:
  # Optional. Specify a server to connect to the registry.
  # Default: No server.
  # May be overridden from the command-line with parameter `--server`.

# --- Worker configuration. Optional. CURRENTLY IGNORED.
workers:
  enabled:
  # A boolean. Specify `true` to launch Synapse with workers.
  # Default: No workers.
  # May be overridden from the command-line with parameter `--workers`.

Debugging

By default, mx-tester has very little in terms of outputs.

mx-tester needs to deal with many tools, each of which has its own logs. To aid you with debugging, mx-tester store all these logs in /tmp/mx-tester/$(YOUR_PROJECT)/logs (or the equivalent directory on your platform). Don't hesitate to look at them :)

The structure roughly looks like:

  • logs/
    • mx-tester/ Logs for the scripts provided in mx-tester.yml
      • up.out, up.log Logs for the up script.
      • run.out, run.log Logs for the run script.
      • modules/ Logs for the build scripts of the modules provided in mx-tester.yml
    • nginx/ If you're running with workers, the nginx load-balancer.
    • workers/ If you're running with workers, the logs for each worker.
    • docker/ The logs for everything running in Docker.
      • build.out, build.log Logs everything that was executed on the guest during mx-tester build step.
      • up-run-down.out, up-run-down.log Logs everything that was executed on the guest during steps mx-tester up, mx-tester run and mx-tester down.

If you wish to look at all the output of these tools during execution, you may prefix your call to mx-tester as follows:

$ RUST_LOG=debug,bollard=error mx-tester run # and/or build, up, down...

Docker notes

Everything is executed with Docker, with the same limitations and abstraction leaks.

Host host.docker.internal should be configured on all platforms, which gives the guest Synapse (and modules) access to the host, should this be needed for tests.

The guest container is running on a network called mx-tester-synapse-$(TAG), where TAG is the Docker tag for the version of Synapse running. By default, that's matrixdotorg/synapse:latest.

Synapse notes

Rate limits

Since version 0.3.3, by default, mx-tester expands considerably the rate limits for Synapse. We assume that this is the preferred behavior for many tests.

If you do NOT wish to expand the rate limits, you need to override them in the homeserver section of mx-tester.yml, e.g.

homeserver:
  # ...
  rc_login: synapse-default # Restores the Synapse default

Dependencies

~41–58MB
~1M SLoC