async def test_monitoring_retry_exceptions(QueueMock, NotifyMock,
mock_taskcluster):
bus = MessageBus()
monitoring = Monitoring(mock_taskcluster, "testqueue", ["pinco@pallino"],
1)
monitoring.register(bus)
await bus.send("testqueue", ("Group1", "Hook1", "Task-exception-retry:2"))
await bus.send("testqueue", ("Group1", "Hook2", "Task-exception-retry:0"))
assert bus.queues["testqueue"].qsize() == 2
monitoring.queue = QueueMock
assert len(monitoring.queue.created_tasks) == 0
monitoring.notify = NotifyMock
# Task exception with 2 retries
await monitoring.check_task()
assert monitoring.stats["Hook1"]["exception"] == ["Task-exception-retry:2"]
assert len(monitoring.queue.created_tasks) == 1
assert bus.queues["testqueue"].qsize() == 2
# The retried task should maintain the original taskGroupId
old_task = await monitoring.queue.task("Task-exception-retry:2")
new_task_id, new_task = monitoring.queue.created_tasks[0]
assert new_task_id != "Task-exception-retry:2"
assert new_task != old_task
assert new_task["taskGroupId"] == old_task["taskGroupId"]
assert new_task["payload"] == old_task["payload"]
assert new_task["created"] != old_task["created"]
# Task exception with 0 retries
# No new task should be created
await monitoring.check_task()
assert monitoring.stats["Hook2"]["exception"] == ["Task-exception-retry:0"]
assert len(monitoring.queue.created_tasks) == 1
assert bus.queues["testqueue"].qsize() == 1
class Events(object):
"""
Listen to pulse events and trigger new code coverage tasks
"""
def __init__(self):
# Create message bus shared amongst process
self.bus = MessageBus()
# Build code coverage workflow
self.workflow = CodeCoverage(
taskcluster_config.secrets["hook_id"],
taskcluster_config.secrets["hook_group_id"],
self.bus,
)
# Setup monitoring for newly created tasks
self.monitoring = Monitoring(
taskcluster_config,
QUEUE_MONITORING,
taskcluster_config.secrets["admins"],
7 * 3600,
)
self.monitoring.register(self.bus)
# Create pulse listener for code coverage
self.pulse = PulseListener(
{
QUEUE_PULSE: [
("exchange/taskcluster-queue/v1/task-group-resolved", ["#"])
]
},
taskcluster_config.secrets["pulse_user"],
taskcluster_config.secrets["pulse_password"],
)
self.pulse.register(self.bus)
def run(self):
consumers = [
# Code coverage main workflow
self.workflow.run(),
# Add monitoring task
self.monitoring.run(),
# Add pulse task
self.pulse.run(),
]
# Run all tasks concurrently
run_tasks(consumers)
async def test_report_all_completed(QueueMock, NotifyMock, mock_taskcluster):
bus = MessageBus()
monitoring = Monitoring(mock_taskcluster, "testqueue", ["pinco@pallino"],
1)
monitoring.register(bus)
await bus.send("testqueue", ("Group1", "Hook1", "Task1-completed"))
await bus.send("testqueue", ("Group1", "Hook1", "Task2-completed"))
assert bus.queues["testqueue"].qsize() == 2
monitoring.queue = QueueMock
monitoring.notify = NotifyMock
await monitoring.check_task()
await monitoring.check_task()
# No email sent, since all tasks were successful.
await monitoring.send_report()
assert NotifyMock.email_obj == {}
assert monitoring.stats == {}
class Events(object):
"""
Listen to HTTP notifications from phabricator and trigger new try jobs
"""
def __init__(self, cache_root):
# Create message bus shared amongst processes
self.bus = MessageBus()
publish = taskcluster_config.secrets["PHABRICATOR"].get(
"publish", False)
# Check the redis support is enabled on Heroku
if heroku.in_dyno():
assert self.bus.redis_enabled is True, "Need Redis on Heroku"
community_config = taskcluster_config.secrets.get(
"taskcluster_community")
test_selection_enabled = taskcluster_config.secrets.get(
"test_selection_enabled", False)
# Run webserver & pulse on web dyno or single instance
if not heroku.in_dyno() or heroku.in_web_dyno():
# Create web server
self.webserver = WebServer(QUEUE_WEB_BUILDS)
self.webserver.register(self.bus)
# Create pulse listener
exchanges = {}
if taskcluster_config.secrets["autoland_enabled"]:
logger.info("Autoland ingestion is enabled")
# autoland ingestion
exchanges[QUEUE_PULSE_AUTOLAND] = [(PULSE_TASK_GROUP_RESOLVED,
["#.gecko-level-3.#"])]
# Create pulse listeners for bugbug test selection task and unit test failures.
if community_config is not None and test_selection_enabled:
exchanges[QUEUE_PULSE_TRY_TASK_END] = [
(PULSE_TASK_COMPLETED, ["#.gecko-level-1.#"]),
(PULSE_TASK_FAILED, ["#.gecko-level-1.#"]),
# https://bugzilla.mozilla.org/show_bug.cgi?id=1599863
# (
# "exchange/taskcluster-queue/v1/task-exception",
# ["#.gecko-level-1.#"],
# ),
]
self.community_pulse = PulseListener(
{
QUEUE_PULSE_BUGBUG_TEST_SELECT: [(
"exchange/taskcluster-queue/v1/task-completed",
["route.project.relman.bugbug.test_select"],
)]
},
taskcluster_config.secrets["communitytc_pulse_user"],
taskcluster_config.secrets["communitytc_pulse_password"],
"communitytc",
)
# Manually register to set queue as redis
self.community_pulse.bus = self.bus
self.bus.add_queue(QUEUE_PULSE_BUGBUG_TEST_SELECT, redis=True)
self.bus.add_queue(QUEUE_PULSE_TRY_TASK_END, redis=True)
else:
self.community_pulse = None
if exchanges:
self.pulse = PulseListener(
exchanges,
taskcluster_config.secrets["pulse_user"],
taskcluster_config.secrets["pulse_password"],
)
# Manually register to set queue as redis
self.pulse.bus = self.bus
self.bus.add_queue(QUEUE_PULSE_AUTOLAND, redis=True)
else:
self.pulse = None
else:
self.bugbug_utils = None
self.webserver = None
self.pulse = None
self.community_pulse = None
logger.info("Skipping webserver, bugbug and pulse consumers")
# Register queues for workers
self.bus.add_queue(QUEUE_PULSE_AUTOLAND, redis=True)
self.bus.add_queue(QUEUE_PULSE_BUGBUG_TEST_SELECT, redis=True)
self.bus.add_queue(QUEUE_PULSE_TRY_TASK_END, redis=True)
self.bus.add_queue(QUEUE_WEB_BUILDS, redis=True)
# Run work processes on worker dyno or single instance
if not heroku.in_dyno() or heroku.in_worker_dyno():
self.workflow = CodeReview(
api_key=taskcluster_config.secrets["PHABRICATOR"]["api_key"],
url=taskcluster_config.secrets["PHABRICATOR"]["url"],
publish=publish,
user_blacklist=taskcluster_config.secrets["user_blacklist"],
)
self.workflow.register(self.bus)
# Build mercurial worker and queue
self.mercurial = MercurialWorker(
QUEUE_MERCURIAL,
QUEUE_MERCURIAL_APPLIED,
repositories=self.workflow.get_repositories(
taskcluster_config.secrets["repositories"],
cache_root,
default_ssh_key=taskcluster_config.secrets["ssh_key"],
),
)
self.mercurial.register(self.bus)
# Setup monitoring for newly created tasks
self.monitoring = Monitoring(
taskcluster_config,
QUEUE_MONITORING,
taskcluster_config.secrets["admins"],
MONITORING_PERIOD,
)
self.monitoring.register(self.bus)
# Setup monitoring for newly created community tasks
if community_config is not None:
self.community_monitoring = Monitoring(
community_taskcluster_config,
QUEUE_MONITORING_COMMUNITY,
taskcluster_config.secrets["admins"],
MONITORING_PERIOD,
)
self.community_monitoring.register(self.bus)
else:
self.community_monitoring = None
self.bugbug_utils = BugbugUtils(self.workflow.api)
self.bugbug_utils.register(self.bus)
else:
self.workflow = None
self.mercurial = None
self.monitoring = None
self.community_monitoring = None
self.bugbug_utils = None
logger.info("Skipping workers consumers")
def run(self):
consumers = []
# Code review main workflow
if self.workflow:
consumers += [
# Process Phabricator build received from webserver
self.bus.run(self.workflow.process_build,
QUEUE_WEB_BUILDS,
sequential=False),
# Publish results on Phabricator
self.bus.run(
self.workflow.publish_results,
QUEUE_PHABRICATOR_RESULTS,
sequential=False,
),
# Trigger autoland tasks
self.bus.run(
self.workflow.trigger_autoland,
QUEUE_PULSE_AUTOLAND,
sequential=False,
),
# Send to phabricator results publication for normal processing and to bugbug for further analysis
self.bus.dispatch(
QUEUE_MERCURIAL_APPLIED,
[QUEUE_PHABRICATOR_RESULTS, QUEUE_BUGBUG_TRY_PUSH],
),
]
if self.bugbug_utils:
consumers += [
self.bus.run(self.bugbug_utils.process_build,
QUEUE_BUGBUG,
sequential=False),
self.bus.run(
self.bugbug_utils.process_push,
QUEUE_BUGBUG_TRY_PUSH,
sequential=False,
),
self.bus.run(
self.bugbug_utils.got_try_task_end,
QUEUE_PULSE_TRY_TASK_END,
sequential=False,
),
self.bus.run(
self.bugbug_utils.got_bugbug_test_select_end,
QUEUE_PULSE_BUGBUG_TEST_SELECT,
sequential=False,
),
]
# Add mercurial task
if self.mercurial:
consumers.append(self.mercurial.run())
# Add monitoring task
if self.monitoring:
consumers.append(self.monitoring.run())
# Add community monitoring task
if self.community_monitoring:
consumers.append(self.community_monitoring.run())
# Add pulse listener for task results.
if self.pulse:
consumers.append(self.pulse.run())
# Add communitytc pulse listener for test selection results.
if self.community_pulse:
consumers.append(self.community_pulse.run())
# Start the web server in its own process
if self.webserver:
self.webserver.start()
if consumers:
# Run all tasks concurrently
run_tasks(consumers)
else:
# Keep the web server process running
asyncio.get_event_loop().run_forever()
# Make sure any pending task is run.
run_tasks(asyncio.Task.all_tasks())
# Stop the webserver when other async processes are stopped
if self.webserver:
self.webserver.stop()
class Events(object):
"""
Listen to HTTP notifications from phabricator and trigger new try jobs
"""
def __init__(self, cache_root):
# Create message bus shared amongst processes
self.bus = MessageBus()
self.workflow = CodeReview(
api_key=taskcluster_config.secrets["PHABRICATOR"]["api_key"],
url=taskcluster_config.secrets["PHABRICATOR"]["url"],
publish=taskcluster_config.secrets["PHABRICATOR"].get(
"publish", False),
risk_analysis_reviewers=taskcluster_config.secrets.get(
"risk_analysis_reviewers", []),
community_config=taskcluster_config.secrets.get(
"taskcluster_community"),
)
self.workflow.register(self.bus)
# Build mercurial worker and queue
self.mercurial = MercurialWorker(
QUEUE_MERCURIAL,
QUEUE_PHABRICATOR_RESULTS,
repositories=self.workflow.get_repositories(
taskcluster_config.secrets["repositories"], cache_root),
)
self.mercurial.register(self.bus)
# Create web server
self.webserver = WebServer(QUEUE_WEB_BUILDS)
self.webserver.register(self.bus)
# Setup monitoring for newly created tasks
self.monitoring = Monitoring(QUEUE_MONITORING,
taskcluster_config.secrets["admins"],
MONITORING_PERIOD)
self.monitoring.register(self.bus)
def run(self):
consumers = [
# Code review main workflow
self.workflow.run(),
# Add mercurial task
self.mercurial.run(),
# Add monitoring task
self.monitoring.run(),
]
# Publish results on Phabricator
if self.workflow.publish:
consumers.append(
self.bus.run(self.workflow.publish_results,
QUEUE_PHABRICATOR_RESULTS))
# Start the web server in its own process
self.webserver.start()
# Run all tasks concurrently
run_tasks(consumers)
# Stop the webserver when other async processes are stopped
self.webserver.stop()
async def test_monitoring(QueueMock, NotifyMock, mock_taskcluster):
bus = MessageBus()
monitoring = Monitoring(mock_taskcluster, "testqueue", ["pinco@pallino"],
1)
monitoring.register(bus)
await bus.send("testqueue", ("Group1", "Hook1", "Task-invalid"))
await bus.send("testqueue", ("Group1", "Hook1", "Task-pending"))
await bus.send("testqueue", ("Group1", "Hook1", "Task1-completed"))
await bus.send("testqueue", ("Group1", "Hook1", "Task2-completed"))
await bus.send("testqueue", ("Group1", "Hook2", "Task-exception"))
await bus.send("testqueue", ("Group2", "Hook1", "Task-failed"))
assert bus.queues["testqueue"].qsize() == 6
monitoring.queue = QueueMock
monitoring.notify = NotifyMock
# No report sent, since we haven't collected any stats yet.
await monitoring.send_report()
assert NotifyMock.email_obj == {}
# Queue throws exception, remove task from queue.
await monitoring.check_task()
assert bus.queues["testqueue"].qsize() == 5
# Task is pending, put it back in the queue.
await monitoring.check_task()
assert bus.queues["testqueue"].qsize() == 5
# No report sent, since we haven't collected any stats yet.
await monitoring.send_report()
assert NotifyMock.email_obj == {}
# Task is completed.
await monitoring.check_task()
assert monitoring.stats["Hook1"]["completed"] == ["Task1-completed"]
assert bus.queues["testqueue"].qsize() == 4
# Another task is completed.
await monitoring.check_task()
assert monitoring.stats["Hook1"]["completed"] == [
"Task1-completed",
"Task2-completed",
]
assert bus.queues["testqueue"].qsize() == 3
# Task exception.
assert len(monitoring.queue.created_tasks) == 0
await monitoring.check_task()
assert monitoring.stats["Hook1"]["exception"] == []
assert monitoring.stats["Hook2"]["exception"] == ["Task-exception"]
# A new task has been retried, replacing the exception
assert len(monitoring.queue.created_tasks) == 1
assert bus.queues["testqueue"].qsize() == 3
# Task failed.
await monitoring.check_task()
assert monitoring.stats["Hook1"]["failed"] == ["Task-failed"]
assert bus.queues["testqueue"].qsize() == 2
# Task is pending, put it back in the queue.
await monitoring.check_task()
assert bus.queues["testqueue"].qsize() == 2
content = """# Hook1 tasks for the last period
## completed
66.67% of all tasks (2/3)
* [Task1-completed](http://taskcluster.test/tasks/Task1-completed)
* [Task2-completed](http://taskcluster.test/tasks/Task2-completed)
## exception
0.00% of all tasks (0/3)
## failed
33.33% of all tasks (1/3)
* [Task-failed](http://taskcluster.test/tasks/Task-failed)
# Hook2 tasks for the last period
## completed
0.00% of all tasks (0/1)
## exception
100.00% of all tasks (1/1)
* [Task-exception](http://taskcluster.test/tasks/Task-exception)
## failed
0.00% of all tasks (0/1)
"""
await monitoring.send_report()
assert NotifyMock.email_obj["address"] == "pinco@pallino"
assert NotifyMock.email_obj["subject"] == "Pulse listener tasks"
assert NotifyMock.email_obj["content"] == content
assert NotifyMock.email_obj["template"] == "fullscreen"
assert monitoring.stats == {}
async def test_monitoring_whiteline_between_failed_and_hook(
QueueMock, NotifyMock, mock_taskcluster):
bus = MessageBus()
monitoring = Monitoring(mock_taskcluster, "testqueue", ["pinco@pallino"],
1)
monitoring.register(bus)
await bus.send("testqueue", ("Group1", "Hook1", "Task-failed"))
await bus.send("testqueue", ("Group1", "Hook2", "Task-failed"))
assert bus.queues["testqueue"].qsize() == 2
monitoring.queue = QueueMock
monitoring.notify = NotifyMock
# Task exception.
await monitoring.check_task()
assert monitoring.stats["Hook1"]["failed"] == ["Task-failed"]
assert bus.queues["testqueue"].qsize() == 1
# Task failed.
await monitoring.check_task()
assert monitoring.stats["Hook2"]["failed"] == ["Task-failed"]
assert bus.queues["testqueue"].qsize() == 0
content = """# Hook1 tasks for the last period
## completed
0.00% of all tasks (0/1)
## exception
0.00% of all tasks (0/1)
## failed
100.00% of all tasks (1/1)
* [Task-failed](http://taskcluster.test/tasks/Task-failed)
# Hook2 tasks for the last period
## completed
0.00% of all tasks (0/1)
## exception
0.00% of all tasks (0/1)
## failed
100.00% of all tasks (1/1)
* [Task-failed](http://taskcluster.test/tasks/Task-failed)"""
await monitoring.send_report()
assert NotifyMock.email_obj["address"] == "pinco@pallino"
assert NotifyMock.email_obj["subject"] == "Pulse listener tasks"
assert NotifyMock.email_obj["content"] == content
assert NotifyMock.email_obj["template"] == "fullscreen"
assert monitoring.stats == {}