Do you want to write some tests using this distributed test setup? Great! Here's what you do.
- Create a branch. Name it something descriptive, but start it with
tests/
please. [Example:tests/data-redundancy
.] - Checkout your branch.
- Create a new assembly in the
/Tests
folder. This can be an NUnit test assembly or simply a console app. - Add Project references to
Core
, as well as any project plugin you'll be using. - Write tests! Use existing tests for inspiration.
- Add new code files to
Tests/CodexTests
- Inherrit from
CodexDistTest
orAutoBootstrapDistTest
. - When using
CodexDistTest
:- You must start your own Codex bootstrap node. You can use
AddCodex(...)
for this. - When you start other Codex nodes with
AddCodex(...)
you can pass the bootstrap node by adding the.WithBootstrapNode(...)
option.
- You must start your own Codex bootstrap node. You can use
- When using
AutoBootstrapDistTest
:- The test-infra creates the bootstrap node for you, and automatically passes it to each Codex node you create in your tests. Handy for keeping your tests clean and to-the-point.
- When using the auto-bootstrap, you have no control over the bootstrap node from your tests. You can't (for example) shut it down during the course of the test. If you need this level of control for your scenario, use the
CodexDistTest
instead.
- If your test needs a long time to run, add the
[UseLongTimeouts]
function attribute. This will greatly increase maximum time-out values for operations like for example uploading and downloading files.
- Add new code files to
Tests/CodexContinuousTests/Tests
- Inherrit from
ContinuousTest
- Define one or more methods and decorate them with the
[TestMoment(...)]
attribute. - The TestMoment takes a number of seconds as argument. Each moment will be executed by the continuous test runner applying the given seconds as delay. (Non-cumulative. So two moments at T:10 will be executed one after another without delay, in this case the order of execution should not be depended upon.)
- Continuous tests automatically receive access to the Codex nodes that the tests are being run against.
- Additionally, Continuous tests can start their own transient Codex nodes and bootstrap them against the persistent nodes.
- You can generate files of random test data by calling
GenerateTestFile(...)
. - You can enable access to the Codex node metrics by adding the option
.EnableMetrics()
. Enabling metrics will make the test-infra download and save all Codex metrics in case of a test failure. (The metrics are stored as CSV, in the same location as the test log file.) - You can enable access to the blockchain marketplace by adding the option
.EnableMarketplace(...)
. - Enabling metrics and/or enabling the marketplace takes extra resources from the test-infra and increases the time needed during Codex node setup. Please don't enable these features unless your tests need them.
- Tip: Codex nodes can be named. Use the option
WithName(...)
and make reading your test logs a little nicer! - Tip: Commit often.
- Once you're happy with your tests, please create a pull-request and ask a Codex core contributor to review your changes.