Run rabbit_ct_hook before cth_styledout so that removing default logger handler works

[ikavgo]

As a follow-up to my GChat thread about removing default logger handler to clean CT stdout, I was looking at
injecting logger config with undefined default handler to ct_run. It is possible but breaks cth_styledout - no nice green things whatsoever. Then I found rabbit_ct_hook which calls redirect_logger_to_ct_logs which in turn calls logger:remove_handler(default) apparently with zero effect! To cut story short - turned out rabbit_ct_hook must run before cth_styledout for remove_handler line to have any effect.

Doesn't work for parallel targets yet -

rabbitmq-server/deps/rabbit/Makefile

Line 227 in 77477de

# @todo We must ensure that the CT_OPTS also apply to ct-master

But for manual local runs like make -C deps/rabbit ct-<> it really shines.

[ikavgo]

Example:
Before: https://github.com/rabbitmq/rabbitmq-server/actions/runs/15231325953/job/42849532519#step:9:1003
After: https://github.com/rabbitmq/rabbitmq-server/actions/runs/15240078576/job/42859060077?pr=13947#step:9:920

Before: https://github.com/rabbitmq/rabbitmq-server/actions/runs/15231325953/job/42849532470#step:9:1539
After: https://github.com/rabbitmq/rabbitmq-server/actions/runs/15240078576/job/42859060076?pr=13947#step:9:891

[ikavgo]

@michaelklishin It works for individual suites, but not for parallel - too much magic. I can add rabbit_ct_hook to other deps if this approach is fine.

[michaelklishin]

@ikavgo I certainly find this output an improvement. We can start with the core (deps/rabbit, deps/rabbit_common) and rabbitmq_management, rabbitmq_mqtt, and then switch more subprojects over time.

[michaelklishin]

Unless @dumbbell @lhoguin have objections, I think we can extend this to a few subprojects with large test suites (see above) and merge this change for main, then see how it goes.

[lhoguin]

Fine with me.

[dumbbell]

Looks good to me as well.

[ikavgo]

cool, I will add to rabbit_common, management and mqtt.

@lhoguin any hints on parallel suites? It looks like it has this "inplace" ct_master right inside the Makefile =). I feel like I plug CT_OPTS somewhere here, but not sure where still...

[lhoguin]

Options need to be converted to their Erlang term equivalent and given to CT from the Erlang script in the Makefile. Not sure about the details.

[ikavgo]

I added crude support for parallel test (ct_master is forked anyway), and replaces some ct:pals with ct:log.

Before: https://github.com/rabbitmq/rabbitmq-server/actions/runs/15240078576/job/42860845449#step:9:13156
After: https://github.com/rabbitmq/rabbitmq-server/actions/runs/15254128554/job/42897594880

[ikavgo]

I gave up on rabbit_ct_hook for rabbit_common. ct_helpers depend on rabbit_common and it is all sorts of ugliness

[michaelklishin]

@ikavgo no worries, we can leave rabbit_common as is for now. Most test suites are not in rabbit_common anyway :)

[ikavgo]

It is green, finally. If looks good, I think I'll add to the rest of the deps and if you are flying I can merge. Thoughts?

[michaelklishin]

Thank you, @ikavgo.