fawind / looper Goto Github PK

This ensures that output logs are not overwritten for different combinations of services and tests.

Only start the single docker service to test when running replay mode.

Do case studies on HPI Schul Cloud or other open source projects.

Do benchmarks for:

• Test speedup
• Ease of integration
• Developer experience

Allow passing in an additional docker-compose.yml to the cli containing user defined overrides.

Wait for container to be up before executing the test command. Possibly using the docker HEALTHCHECK option.

Read cli params from local config file (e.g. docker-test.yml).

Using spf13/viper is probably a good idea.

Add cli option to pass env variables to test (and maybe docker?) command.

e.g. ./docker-test [...] -env 'var1=val1'

Add option to override services with mock versions. E.g. by providing a mapping of service-name to mock docker image.

Handle stateful requests. Example:

1. Get: /hitcount -> 1
2. Get: /hitcount -> 2

Options:

• Check if mitmdump already handles such cases
• Add unique request id to url param or header
  • Ids have to be ordered in order to match them in record and replay mode?

• Allow the proxy to capture HTTPS traffic.
• mitm proxy seems to already support this: docs
• Set up a working concept with notes-service using nginx

• Allow starting just the mitm proxy to capture test traffic of services not running in docker.
• Clients are given the url of the proxy and connect manually.
• Should probably done as a cli-flag that starts just the mitm proxy.