Unstarted jobs should always be cleaned up (#4604)

This commit is contained in:
Mike Degatano 2023-10-09 05:57:04 -04:00 committed by GitHub
parent f9840306a0
commit ace58ba735
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 164 additions and 88 deletions

View File

@ -86,7 +86,7 @@ class SupervisorJob:
} }
@contextmanager @contextmanager
def start(self, *, on_done: Callable[["SupervisorJob"], None] | None = None): def start(self):
"""Start the job in the current task. """Start the job in the current task.
This can only be called if the parent ID matches the job running in the current task. This can only be called if the parent ID matches the job running in the current task.
@ -107,8 +107,6 @@ class SupervisorJob:
self.done = True self.done = True
if token: if token:
_CURRENT_JOB.reset(token) _CURRENT_JOB.reset(token)
if on_done:
on_done(self)
class JobManager(FileConfiguration, CoreSysAttributes): class JobManager(FileConfiguration, CoreSysAttributes):

View File

@ -201,6 +201,7 @@ class Job(CoreSysAttributes):
internal=self._internal, internal=self._internal,
) )
try:
# Handle condition # Handle condition
if self.conditions: if self.conditions:
try: try:
@ -211,14 +212,19 @@ class Job(CoreSysAttributes):
return self._handle_job_condition_exception(err) return self._handle_job_condition_exception(err)
# Handle exection limits # Handle exection limits
if self.limit in (JobExecutionLimit.SINGLE_WAIT, JobExecutionLimit.ONCE): if self.limit in (
JobExecutionLimit.SINGLE_WAIT,
JobExecutionLimit.ONCE,
):
await self._acquire_exection_limit() await self._acquire_exection_limit()
elif self.limit in ( elif self.limit in (
JobExecutionLimit.GROUP_ONCE, JobExecutionLimit.GROUP_ONCE,
JobExecutionLimit.GROUP_WAIT, JobExecutionLimit.GROUP_WAIT,
): ):
try: try:
await obj.acquire(job, self.limit == JobExecutionLimit.GROUP_WAIT) await obj.acquire(
job, self.limit == JobExecutionLimit.GROUP_WAIT
)
except JobGroupExecutionLimitExceeded as err: except JobGroupExecutionLimitExceeded as err:
if self.on_condition: if self.on_condition:
raise self.on_condition(str(err)) from err raise self.on_condition(str(err)) from err
@ -244,26 +250,35 @@ class Job(CoreSysAttributes):
JobExecutionLimit.GROUP_THROTTLE_RATE_LIMIT, JobExecutionLimit.GROUP_THROTTLE_RATE_LIMIT,
): ):
# Only reprocess array when necessary (at limit) # Only reprocess array when necessary (at limit)
if len(self.rate_limited_calls(group_name)) >= self.throttle_max_calls: if (
len(self.rate_limited_calls(group_name))
>= self.throttle_max_calls
):
self.set_rate_limited_calls( self.set_rate_limited_calls(
[ [
call call
for call in self.rate_limited_calls(group_name) for call in self.rate_limited_calls(group_name)
if call > datetime.now() - self.throttle_period(group_name) if call
> datetime.now() - self.throttle_period(group_name)
], ],
group_name, group_name,
) )
if len(self.rate_limited_calls(group_name)) >= self.throttle_max_calls: if (
len(self.rate_limited_calls(group_name))
>= self.throttle_max_calls
):
on_condition = ( on_condition = (
JobException if self.on_condition is None else self.on_condition JobException
if self.on_condition is None
else self.on_condition
) )
raise on_condition( raise on_condition(
f"Rate limit exceeded, more then {self.throttle_max_calls} calls in {self.throttle_period(group_name)}", f"Rate limit exceeded, more then {self.throttle_max_calls} calls in {self.throttle_period(group_name)}",
) )
# Execute Job # Execute Job
with job.start(on_done=self.sys_jobs.remove_job if self.cleanup else None): with job.start():
try: try:
self.set_last_call(datetime.now(), group_name) self.set_last_call(datetime.now(), group_name)
if self.rate_limited_calls(group_name) is not None: if self.rate_limited_calls(group_name) is not None:
@ -291,6 +306,11 @@ class Job(CoreSysAttributes):
): ):
obj.release() obj.release()
# Jobs that weren't started are always cleaned up. Also clean up done jobs if required
finally:
if job.done is None or self.cleanup:
self.sys_jobs.remove_job(job)
return wrapper return wrapper
@staticmethod @staticmethod

View File

@ -1044,3 +1044,61 @@ async def test_job_starting_separate_task(coresys: CoreSys):
await test.job_await() await test.job_await()
await test.job_release() await test.job_release()
await task await task
async def test_job_always_removed_on_check_failure(coresys: CoreSys):
"""Test that the job instance is always removed if the condition or limit check fails."""
class TestClass:
"""Test class."""
event = asyncio.Event()
limit_job: Job | None = None
def __init__(self, coresys: CoreSys) -> None:
"""Initialize object."""
self.coresys = coresys
@Job(
name="test_job_always_removed_on_check_failure_condition",
conditions=[JobCondition.HAOS],
on_condition=JobException,
cleanup=False,
)
async def condition_check(self):
"""Job that will fail a condition check."""
raise AssertionError("should not run")
@Job(
name="test_job_always_removed_on_check_failure_limit",
limit=JobExecutionLimit.ONCE,
cleanup=False,
)
async def limit_check(self):
"""Job that can fail a limit check."""
self.limit_job = self.coresys.jobs.current
await self.event.wait()
def release_limit_check(self):
"""Release the limit check job."""
self.event.set()
test = TestClass(coresys)
with pytest.raises(JobException):
await test.condition_check()
assert coresys.jobs.jobs == []
task = coresys.create_task(test.limit_check())
await asyncio.sleep(0)
assert (job := test.limit_job)
with pytest.raises(JobException):
await test.limit_check()
assert test.limit_job == job
assert coresys.jobs.jobs == [job]
test.release_limit_check()
await task
assert job.done
assert coresys.jobs.jobs == [job]