Eight scenarios in, the same few things keep deciding whether a scenario is
worth having, and none of them were obvious beforehand:

- Falsify every assertion, not just enough to see red. One control asserted that
  a component emitted no DEBUG records from a module and passed just as happily
  with that module set to `debug`, because it emits none on a first run either
  way. Green for the wrong reason, and only breaking it deliberately showed that.
- `ActiveState == active` with `NRestarts == 0` can both hold while a component
  is completely broken, when it catches its errors and retries rather than
  exiting. Two of the eight behave that way.
- Startup lines are the oldest in the journal, so tailing loses them. Grep, and
  strip ANSI first.
- Parse rendered configuration and assert on structure, so a value under the
  wrong key cannot pass.
- `molecule converge` on a running instance does not restart the container, so a
  falsification can pass for that reason alone.

Also documents that a role's scenario obliges it to join the automerge list, and
that prek enforces the two staying in step.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SEH3vxYSQ5SV4N5z61eyGT

`group_vars/matrix_servers` selects postgres whenever postgres is enabled, which
is the default, so postgres is what essentially every deployment runs. The
scenarios were testing sqlite - a path almost nobody is on.

How little that path is used is not a guess: the mautrix-meta bridges could not
start at all under sqlite, and nobody reported it. Testing the engine users are
actually on is worth more than keeping coverage of the one they are not, so no
scenario is left behind on sqlite.

Four of the eight scenarios have a database and are converted; the other four
have none and are untouched.

molecule-shared/tasks/postgres.yml stands Postgres up on the scenario's network,
with the data directory on a tmpfs since it is thrown away with the container.
The image is pinned at the major the postgres role deploys to new installations
and left to Renovate: when a new major lands, the PR bumping that pin runs every
scenario against it, which is the earliest warning we get that a component does
not cope.

Each scenario gives its database and user names that differ from the role's
defaults, so the component reaching the database proves the role built its
connection string out of them. The assertions moved from "a file appeared at the
path we configured" to "these tables exist", which is strictly stronger: tables
can only appear once the component has resolved the hostname, authenticated with
the credentials the role rendered, and run its migrations to completion.

Costs about 10 seconds per affected scenario (115s to 125s locally for
mautrix-whatsapp), on jobs that run in parallel.

Gotcha worth recording: since Postgres 18 the image puts PGDATA in a versioned
subdirectory and refuses to start if it finds a mount at the old
/var/lib/postgresql/data, so the tmpfs is mounted at /var/lib/postgresql.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SEH3vxYSQ5SV4N5z61eyGT

`just molecule` leaves a shared virtualenv of over 500 MB under var/, plus a
~7 MB Ansible home per role that has a scenario. Neither is reclaimed by
anything today.

`--idle-days N` limits it to what has not been touched recently, so it can be
run unattended without taking the cache out from under a scenario being worked
on right now.

The two directories are named explicitly rather than globbed: var/ holds other
things and must never be removed wholesale.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SEH3vxYSQ5SV4N5z61eyGT

Scenarios install their Galaxy dependencies with `force: true`, so two roles
running at once re-extract the same collections and roles into ~/.ansible and
pull them out from under each other mid-play. It surfaces as a collection that
was working moments earlier going missing:

  the connection plugin 'community.docker.docker' was not found

Found while running five scenarios in parallel, where it cost a run.

ANSIBLE_HOME relocates both `collections/` and `roles/`, so one variable covers
both halves; the scenarios' ANSIBLE_ROLES_PATH workaround now follows it rather
than hardcoding ~/.ansible/roles. Left alone if already set, and unset in CI,
where each role runs in its own job and has nothing to collide with.

Verified by removing var/molecule-ansible-home entirely and running
matrix-alertmanager-receiver from cold: green through idempotence, with the
collections and roles landing under the per-role directory - which also shows
nothing was quietly relying on the shared ~/.ansible being populated.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SEH3vxYSQ5SV4N5z61eyGT

Molecule checks for a requirements file at the scenario's default path
before it installs anything, and warns "Missing roles requirements file"
when it is absent - so `requirements-file` pointing elsewhere was silently
ignored and the galaxy roles were never installed. CI failed on a missing
ansible-role-docker; the local run passed only because those roles were
already in ~/.ansible/roles from earlier work, which is exactly the kind of
difference a clean CI run exists to expose.

A symlink satisfies Molecule's check while keeping one copy of the pins.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Three things that would not have scaled to 70 roles:

- The Python and Ansible dependency pins were about to be copied into every
  role. They now live once in molecule-shared/, which scenarios reference
  relatively, so they cannot drift apart.
- The helper container images used for probing were hardcoded inline. They
  are pinned once in molecule-shared/vars.yml, carry `# renovate:`
  annotations, and a custom manager in .github/renovate.json keeps them
  current - verified with a local Renovate dry run, which offers
  curl 8.11.1 -> 8.21.0 and python 3.13 -> 3.14-alpine. Seventy invisible
  hardcodes is the blindness class we have been removing elsewhere.
- Running a scenario meant knowing the venv and cd incantation. `just
  molecule <role>` does it, and with no argument lists the roles that have
  a scenario.

Molecule is deliberately not wired into prek: a run takes minutes, pulls
images and needs Docker, which is fine on request and not fine per commit.

docs/molecule-testing.md covers how to run and write these, including the
four things a role here needs that a standalone role does not. AGENTS.md
points at it rather than carrying the detail.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>