1
0
mirror of https://blitiri.com.ar/repos/chasquid synced 2025-12-17 14:37:02 +00:00
Files
go-chasquid-smtp/test
Alberto Bertogli d3c18aa471 modules: Update spf to 1.3.0
This patch updates the dependency on blitiri.com.ar/go/spf from v1.2.0
to v1.3.0, which includes a few bug fixes.

There are no code changes needed, just some minor adjustment to the
tests due to error strings changing.

The go.mod "go" keyword is also bumped up to 1.15 since it's the minimum
supported version since commit e444fe1f (2021-10-05).
2022-02-27 11:30:34 +00:00
..
2022-02-27 11:30:34 +00:00
2021-07-21 02:06:20 +01:00
2018-03-26 01:58:47 +01:00
2020-08-21 12:07:33 +01:00
2021-07-21 02:06:20 +01:00
2018-03-26 01:58:47 +01:00

Testing

Go tests

All Go packages have their own test suite, which provides easy and portable tests with decent enough coverage.

Integration tests

In the test/ directory there is a set of end to end integration tests, written usually in a combination of bash and Python 3.

They're not expected to be portable, as that gets impractical very quickly, but should be usable in most Linux environments.

They provide critical coverage and integration tests for real life scenarios, as well as interactions with other software (like Exim or Dovecot).

Dependencies

The tests depend on the following things being installed on the system (listed as Debian package, for consistency):

  • msmtp
  • util-linux (for /usr/bin/setsid)

Some individual tests have additional dependencies, and the tests are skipped if the dependencies are not found:

  • t-02-exim Exim interaction tests:
    • gettext-base (for /usr/bin/envsubst)
    • The exim binary available somewhere, but it doesn't have to be installed. There's a script get-exim4-debian.sh to get it from the archives.
  • t-11-dovecot Dovecot interaction tests:
    • dovecot
  • t-15-driusan_dkim DKIM integration tests:
    • The dkimsign dkimverify dkimkeygen binaries, from driusan/dkim (no Debian package yet).
  • t-18-haproxy HAProxy integration tests:
    • haproxy

For some tests, python >= 3.5 is required; they will be skipped if it's not available.

Most tests depend on the $HOSTALIASES environment variable being functional, and will be skipped if it isn't. This works by default in most Linux systems, but note that the use of systemd-resolved can prevent it from working properly.

Stress tests

Also in the test/ directory there is a set of stress tests, which generate load against chasquid to measure performance and resource consumption.

While they are not exhaustive, they are useful to catch regressions and track improvements on the main code paths.

Fuzz tests

Some Go packages also have instrumentation to run fuzz testing against them, with the go-fuzz tool.

This is critical for packages that handle sensitive user input, such as authentication encoding, aliases files, or username normalization.

They are implemented by a fuzz.go file within their respective Go packages.

Command-line tool tests

Each command-line tool has their own set of tests, see the test.sh file on their corresponding directories.

Docker

The test/Dockerfile can be used to set up a suitable isolated environment to run the integration and stress tests.

This is very useful for automated tests, or running the integration tests in constrained or non supported environments.

Automated tests

There are two sets of automated tests which are run on every commit to upstream, and weekly:

  • GitLab CI, configured in the .gitlab-ci.yml file, runs the Go tests and the integration tests (using docker).
    The integration tests are run twice: once against the dependencies listed in go.mod, and once against the latest version of the dependencies. It also builds the public Docker images.

  • Cirrus CI, configured in the .cirrus.yml file, runs Go tests on FreeBSD, and a comprehensive linter.

Coverage

The test/cover.sh script runs the integration tests in coverage mode, and produces a code coverage report in HTML format, for ease of analysis.

Unfortunately, exiting with any of the Fatal functions does not save coverage output. Those paths are very important to test, but don't expect to see them reflected in the coverage report for now.

The target is to keep coverage of the chasquid binary above 90%.