RFR: 7539: JDP Tests failing in VPN environments

[Christoph Langer]

On Fri, 21 Jan 2022 15:12:30 GMT, Marcus Hirt <hirt at openjdk.org> wrote:

> I think it would be better to have the test provide information about how to properly configure the machine to properly run the tests than to automatically skip the tests. Perhaps we could also add a system property check or something to make it easy to skip the test from the command line, if someone truly can't get it to work. But then it should be a conscious choice, not something happening automatically to avoid having to fix the setup.
> 
> Here is for example one possible scenario if we go with automated skipping:
> 
> 1. Images used for our automated testing change their configuration, no longer providing a route for local multicast.
> 2. Tests are automatically skipped.
> 3. JDP changes and breaks.
> 4. We are never the wiser (until some JDP using person complains post release).
> 
> We shouldn't automatically skip tests without a very strong reason. It's better to inform the runner of the test how to fix their config.

I agree - we should catch the failure and then log something like:
This test failed. Please check if your multicast is working or you are running in a VPN configuration. If this can't be fixed, build with `mvn -DskipJDPMulticastTests=true ...`.

And then check for this property when deciding whether to run the test.

-------------

PR: https://git.openjdk.java.net/jmc/pull/369