Add agent/collector mode toggle to test setups

[unflxw]

Part of appsignal/appsignal-ruby#1515.

AppSignal integrations are gaining a "collector mode", in which the
integration sets up OpenTelemetry itself and ships data to AppSignal
through a local AppSignal collector. Rather than copying each setup with
a suffix, this makes a single setup toggle-able between "agent mode" (the
default) and "collector mode" with a mode= parameter.

It wires up the toggle, converts a representative subset to prove it out,
removes the now-redundant -integration-collector setup, renames the
manual-OpenTelemetry -collector setups to -opentelemetry to avoid
ambiguity, and then rolls the toggle out to the rest of the Ruby 3.1+ and
Python agent setups.

Add agent/collector mode toggle to rake tasks

Test setups can now be run in either "agent mode" (the native
integration, the default) or "collector mode" (the integration sends
data to AppSignal through a local AppSignal collector over
OpenTelemetry), selected with a mode= parameter.

A setup supports collector mode when it ships a
docker-compose.collector.yml; asking for an unsupported mode errors,
and the chosen mode is printed at the start of the command output. The
shared AppSignal collector service lives in support/collector so per-app
collector files can include it.

Also fix the README bot-app detection, which matched an inline service
that is now pulled in via an include, so regeneration is faithful again.

Add collector mode to a representative set of setups

Convert ruby/grape, ruby/rack, ruby/rails7-postgres, ruby/rails8-sidekiq
and python/django5-celery to the agent/collector toggle. Each is split
into a shared compose file plus per-mode files: the agent file keeps the
existing app name, the collector file adds the AppSignal collector,
points the integration at it, and reports under a -collector app name
so the two modes stay separate in AppSignal.

The Ruby apps gain the optional OpenTelemetry gems collector mode needs;
they are only loaded when collector mode is active.

Remove redundant django5-celery-integration-collector

The django5-celery setup now supports collector mode through the mode
toggle, so the separate integration-collector setup is no longer needed.
Its only app-level difference, a /logs endpoint that emits log messages,
is folded into django5-celery.

Rename -collector setups to -opentelemetry

These setups wire up an OpenTelemetry stack by hand to send data to
AppSignal through the collector. Now that integrations can do this
themselves in collector mode, "-collector" is ambiguous: rename them to
"-opentelemetry" to distinguish the manual setups from the integration's
collector mode.

Renames the directories and updates the app names, image names and test
labels. The collector service and its host name are unchanged.

Add collector mode to Ruby 3.1+ agent setups

Roll the agent/collector toggle out to the remaining Ruby 3.1+ agent
setups, following the pattern already used for the initial subset: a
shared compose file per app, included by the agent and collector files,
with a mode-aware app name and the OpenTelemetry gems needed for
collector mode.

Add collector mode to Python agent setups

Apply the same agent/collector toggle to the remaining Python agent
setups. Python needs no dependency changes; collector mode is enabled
purely by the collector compose file and the mode-aware app name.

[backlog-helper]

✔️ All good!

New issue guide | Backlog management | Rules | Feedback

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[backlog-helper]

• This Pull Request needs more reviews. @tombruijn @lipskis - (More info)

---

This is a message from the daily scheduled checks.

New issue guide | Backlog management | Rules | Feedback