systemd unit file fixes and config changes

AGENTS.md

@@ -0,0 +1,16 @@
+# Repository Guidelines
+
+## Project Structure & Module Organization
+The core Python package lives in `mozilla_bitbar_devicepool/`, with vendor-specific logic in `bitbar/` and `lambdatest/`. Supporting CLI entry points are defined in `pyproject.toml` and mirrored as helper scripts in `bin/`. Shared utils and report generators sit under `util/` and `device_group_report*.py`, while environment and service configuration lives in `config/` and `service/`. Tests reside in `mozilla_bitbar_devicepool/test/`, alongside fixture data in `test_data/`.
+
+## Build, Test, and Development Commands
+Run `poetry install --with=dev` to sync dependencies, then `poetry shell` for an interactive environment. Use `pre-commit install` once per clone to enable formatting and lint hooks. Execute unit tests with `pytest` or the coverage variant `pytest --cov --cov-report=term`. CLI entry points run via `poetry run`, e.g., `poetry run mld` for LambdaTest device runs or `poetry run dgr` for Bitbar reports.
+
+## Coding Style & Naming Conventions
+Stick to Python 3.9+ with 4-space indentation and descriptive, snake_case names for modules, functions, and variables. Formatting and imports are enforced by `ruff-format` with a 120-character line limit; structural linting runs via `ruff check --select I --fix` and `pylint` in CI/pre-commit. Configuration files use YAML or TOML—mirror existing key naming and comment styles.
+
+## Testing Guidelines
+Place new tests in `mozilla_bitbar_devicepool/test/` following the `<module>_test.py` pattern and reuse helpers in `util_test.py`. Prefer `pytest` fixtures over ad-hoc setup; store serialized inputs in `test_data/`. Validate coverage locally with `pytest --cov --cov-report=html` and ensure critical branches that touch vendor APIs are exercised via mocks.
+
+## Commit & Pull Request Guidelines
+Commit messages are short (≤72 characters), lowercase, and often scoped (`lt status: refresh cache`). Group related edits per commit and reference Bugzilla or GitHub IDs when applicable. Pull requests should describe vendor impact, outline testing (`pytest`, manual device run), and include screenshots or log snippets when UI or report output changes. Link configuration updates to rollout plans so reviewers can validate credentials and scheduling.

bin/start_mbd_new_server.sh

@@ -4,4 +4,4 @@ set -e
 set -x
 
 . ./bitbar_env-v3-server.sh
-mbd start-test-run-manager -b config/config-v3-server.yml --update-bitbar
+poetry run mbd start-test-run-manager -b config/config-v3-server.yml --update-bitbar

config/lambdatest.yml

@@ -48,7 +48,6 @@ device_groups:
     R5CXC1AHV4M:  # 9/10/25: recabled and verified by jmaher
     R5CXC1ASA3P:  # 9/11/25: recabled and verified by jmaher
     R5CXC1AP2KT:  # 9/11/25: recabled and verified by jmaher
-    R5CXC1HZK0W:  # 9/11/25: recabled and verified by jmaher
     R5CXC1ASA2E:  # 9/11/25: recabled and verified by jmaher
     R5CXC1ARM0A:  # 9/11/25: recabled and verified by jmaher
     R5CXC1AMNFY:  # 9/15/25: recabled and verified by jmaher
@@ -92,9 +91,10 @@ device_groups:
     R5CXC1AHXYD:  # 10/20/25: tested by joel, power meter good
     R5CXC1ARZDN:  # 10/20/25: tested by joel, power meter good
     R5CXC1HZA6V:  # 10/20/25: tested by joel, power meter good
+    RZCY10LGB6W:  # 2/23/26: had bad rotation, rotation code landed on master
+    RZCY2011M7N:  # 2/23/26: had bad rotation, rotation code landed on master
   a55-alpha:
-    RZCY2011M7N:  # 11/6/25: bad rotation, testing in alpha
-    RZCY10LGB6W:  # 11/6/25: bad rotation, testing in alpha
+    R5CXC1HZK0W:  # now on android 15, move to alpha for testing
   a55-android16:
     R5CXC1ASA0L:  # 12/1/25: lt is attempting to upgrade to android 16, had bad power meter
     RZCY203N75Z:  # 12/1/25: lt is attempting to upgrade to android 16, had bad power meter

mozilla_bitbar_devicepool/configuration_device_mover.py

@@ -162,7 +162,7 @@ def move_devices_from_any_pool(
 
         # Initialize target group if it's None/empty
         if device_groups[target_group] is None:
-            device_groups[target_group] = {}
+            device_groups[target_group] = CommentedMap()
 
         # Track results
         results = {"moved": [], "not_found": [], "already_in_target": [], "errors": []}

service/bitbar.service

@@ -1,12 +1,15 @@
 [Unit]
 Description=Bitbar Testrun manager service
+# retry forever, no rate limiting
+StartLimitBurst=0
 
 [Service]
 Type=simple
 ExecStart=/bin/bash /home/bitbar/mozilla-bitbar-devicepool/bin/start_android_hardware_testing.sh
 # disable execstop, let systemd track the pid and manage shutdown
 #ExecStop=/bin/bash /home/bitbar/mozilla-bitbar-devicepool/bin/stop_android_hardware_testing.sh
 Restart=always
+RestartSec=30s
 # address intermittent hangs by restarting hourly (mega-punt), haven't been able to find deadlocks/source.
 RuntimeMaxSec=1h
 WorkingDirectory=/home/bitbar/mozilla-bitbar-devicepool

service/lambdatest.service

@@ -1,12 +1,15 @@
 [Unit]
 Description=Lambdatest Testrun manager service
+# retry forever, no rate limiting
+StartLimitBurst=0
 
 [Service]
 Type=simple
 ExecStart=/bin/bash /home/bitbar/mozilla-bitbar-devicepool/bin/start_ltdp.sh
 # disable execstop, let systemd track the pid and manage shutdown
 #ExecStop=/bin/bash /home/bitbar/mozilla-bitbar-devicepool/bin/stop_ltdp.sh
 Restart=always
+RestartSec=30s
 # restart process 24 hours to avoid any funk building up
 #  - disabled to find long-term bugs
 #RuntimeMaxSec=24h