Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

tasks: add support for both 'jobs-android' and device specific subqueues #4013

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 24 additions & 1 deletion src/clusterfuzz/_internal/base/tasks.py
Original file line number Diff line number Diff line change
Expand Up @@ -122,6 +122,11 @@ def default_queue_suffix():
return queue_suffix_for_platform(environment.platform())


def generic_android_queue(prefix=JOBS_PREFIX):
"""Get the generic 'android' queue that is not tied to a specific device."""
return prefix + queue_suffix_for_platform(environment.platform())


def regular_queue(prefix=JOBS_PREFIX):
"""Get the regular jobs queue."""
return prefix + default_queue_suffix()
Expand Down Expand Up @@ -190,6 +195,21 @@ def get_regular_task(queue=None):
return task


def get_android_regular_task():
"""Get an android regular task."""
if not environment.get_value('CF_ALLOWS_LISTENING_GENERIC_QUEUE') or \
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As discussed in meet, let's make this default as opposed to introducing new env var

not environment.get_value('QUEUE_OVERRIDE'):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't believe we set the QUEUE_OVERRIDE var anywhere in our tests. Unless it's done via some environment init, should this be OS_OVERRIDE?

return get_regular_task()

# We first try to grab a task on the regular default queue, which is the
# device specific subqueue.
task = get_regular_task(queue=regular_queue())
if not task:
# We then try the generic queue.
task = get_regular_task(queue=generic_android_queue())
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm concerned that pulling from the generic queue will lead to incompatible testcases being picked up (like the error log shows below).

platform (android:coral_hwasan:master) does not match with ours (android:redfin_hwasan:master), exiting-

Is that right? If so, the testcase ("task") will be re-added to the top of the queue and continue in a cycle, meaning the bot will never move on to fuzzing tasks, which only occur when no testcases are left.

return task


def get_machine_template_for_queue(queue_name):
"""Gets the machine template for the instance used to execute a task from
|queue_name|. This will be used by tworkers to schedule the appropriate
Expand Down Expand Up @@ -316,7 +336,10 @@ def get_task():
if task:
return task

task = get_regular_task()
if environment.is_android():
task = get_android_regular_task()
else:
task = get_regular_task()
if task:
# Log the task details for debug purposes.
logs.log(f'Got task with cmd {task.command} args {task.argument} '
Expand Down
19 changes: 19 additions & 0 deletions src/clusterfuzz/_internal/bot/tasks/commands.py
Original file line number Diff line number Diff line change
Expand Up @@ -73,6 +73,19 @@ class AlreadyRunningError(Error):
"""Exception raised for a task that is already running on another bot."""


def get_queue_for_testcase(testcase_platform_id, current_platform_id):
"""Gets the queue where to schedule the task on. Returns None if the default
regular queue should be used.
"""
if environment.is_android(testcase_platform_id.upper()):
# FIXME: Handle these imports in a cleaner way.
from clusterfuzz._internal.platforms import android
if android.util.should_schedule_on_generic_queue(testcase_platform_id,
current_platform_id):
return tasks.generic_android_queue()
return None


def cleanup_task_state():
"""Cleans state before and after a task is executed."""
# Cleanup stale processes.
Expand Down Expand Up @@ -339,10 +352,16 @@ def process_command_impl(task_name,
logs.log('Testcase %d platform (%s) does not match with ours\
(%s), exiting' % (testcase.key.id(), testcase_platform_id,
current_platform_id))
# If this happens on android, we need to make sure we're not
# scheduling this task on a subqueue, since we know the current
# platform doesn't match.
queue = get_queue_for_testcase(testcase_platform_id,
current_platform_id)
tasks.add_task(
task_name,
task_argument,
job_name,
queue=queue,
wait_time=utils.random_number(1, TASK_RETRY_WAIT_LIMIT))
return None

Expand Down
21 changes: 21 additions & 0 deletions src/clusterfuzz/_internal/platforms/android/util.py
Original file line number Diff line number Diff line change
Expand Up @@ -82,3 +82,24 @@ def can_testcase_run_on_platform(testcase_platform_id, current_platform_id):
return True

return False


def should_schedule_on_generic_queue(testcase_platform_id, current_platform_id):
"""Whether the testcase should be scheduled on the generic queue or the
default queue."""
if not environment.get_value('QUEUE_OVERRIDE'):
return False

testcase_fields = testcase_platform_id.split(':')
current_platform_fields = current_platform_id.split(':')

if len(testcase_fields) != 3 or len(current_platform_fields) != 3:
return False

testcase_codename = testcase_fields[1].split('_')
current_platform_codename = current_platform_fields.split('_')

if len(testcase_codename) != 2 or len(current_platform_codename) != 2:
return False

return testcase_codename[0] != testcase_codename[1]
56 changes: 56 additions & 0 deletions src/clusterfuzz/_internal/tests/appengine/common/tasks_test.py
Original file line number Diff line number Diff line change
Expand Up @@ -381,3 +381,59 @@ def test_not_progression(self):
messages = self.client.pull_from_subscription(
self.subscription, max_messages=1)
self.assertEqual(0, len(messages))


@test_utils.integration
@test_utils.with_cloud_emulators('datastore', 'pubsub')
class GetTaskAndroidTest(unittest.TestCase):
"""GetTask tests."""

def setUp(self):
helpers.patch_environ(self)
helpers.patch(self, [
'clusterfuzz._internal.base.persistent_cache.get_value',
'clusterfuzz._internal.base.persistent_cache.set_value',
'clusterfuzz._internal.base.utils.utcnow',
'time.sleep',
])

self.mock.get_value.return_value = None
self.mock.sleep.return_value = None
data_types.Job(name='job').put()

client = pubsub.PubSubClient()
topic = pubsub.topic_name('test-clusterfuzz', 'jobs-android')
client.create_topic(topic)
client.create_subscription(
pubsub.subscription_name('test-clusterfuzz', 'jobs-android'), topic)

topic = pubsub.topic_name('test-clusterfuzz', 'jobs-android-pixel8')
client.create_topic(topic)
client.create_subscription(
pubsub.subscription_name('test-clusterfuzz', 'jobs-android-pixel8'),
topic)

self.mock.utcnow.return_value = test_utils.CURRENT_TIME.replace(
microsecond=0)
environment.set_value('OS_OVERRIDE', 'ANDROID')

def test_device_specific_task(self):
"""Test that device specific tasks are correctly scheduled and pulled."""
environment.set_value('QUEUE_OVERRIDE', 'ANDROID:PIXEL8')
tasks.add_task('test', 'normal', 'job', wait_time=0)
task = tasks.get_task()
self.assertEqual('test', task.command)
self.assertEqual('normal', task.argument)
self.assertEqual('job', task.job)
self.assertEqual('test normal job', task.payload())

def test_generic_task_is_pulled(self):
"""Test that generic tasks are pulled when QUEUE_OVERRIDE is set."""
environment.set_value('CF_ALLOWS_LISTENING_GENERIC_QUEUE', 1)
environment.set_value('QUEUE_OVERRIDE', 'ANDROID:PIXEL8')
tasks.add_task('test', 'normal', 'job', queue='jobs-android', wait_time=0)
task = tasks.get_task()
self.assertEqual('test', task.command)
self.assertEqual('normal', task.argument)
self.assertEqual('job', task.job)
self.assertEqual('test normal job', task.payload())
Loading