I am using a Flask app with the gunicorn server and the gevent worker class, which according to the gunicorn documentation is an asynchronous worker. However, when I launch gunicorn with a single worker and try to make a long request (I added sleep(10) in the route function, but in reality this also happens when processing large uploads), I can't make any request until the previous one is finished. It behaves as is it is a synchronous worker, one request at a time.
Is this the normal behavior? Am I missing something about synchronous vs asynchronous workers?
If you don't monkey-patch sleep (or use gevent's non-blocking version of sleep) then a worker that blocks blocks the entire event loop.
Either call gevent.monkey.patch_all (or more specifically gevent.monkey.patch_time) or replace your call to time.sleep with gevent.sleep
Related
We have a web app made with pyramid and served through gunicorn+nginx. It works with 8 worker threads/processes
We needed to jobs, we have chosen apscheduler. here is how we launch it
from apscheduler.events import EVENT_JOB_EXECUTED, EVENT_JOB_ERROR
from apscheduler.scheduler import Scheduler
rerun_monitor = Scheduler()
rerun_monitor.start()
rerun_monitor.add_interval_job(job_to_be_run,\
seconds=JOB_INTERVAL)
The issue is that all the worker processes of gunicorn pick the scheduler up. We tried implementing a file lock but it does not seem like a good enough solution. What would be the best way to make sure at any given time only one of the worker process picks the scheduled event up and no other thread picks it up till next JOB_INTERVAL?
The solution needs to work even with mod_wsgi in case we decide to switch to apache2+modwsgi later. It needs to work with single process development server which is waitress.
Update from the bounty sponsor
I'm facing the same issue described by the OP, just with a Django app. I'm mostly sure adding this detail won't change much if the original question. For this reason, and to gain a bit more of visibility, I also tagged this question with django.
Because Gunicorn is starting with 8 workers (in your example), this forks the app 8 times into 8 processes. These 8 processes are forked from the Master process, which monitors each of their status & has the ability to add/remove workers.
Each process gets a copy of your APScheduler object, which initially is an exact copy of your Master processes' APScheduler. This results in each "nth" worker (process) executing each job a total of "n" times.
A hack around this is to run gunicorn with the following options:
env/bin/gunicorn module_containing_app:app -b 0.0.0.0:8080 --workers 3 --preload
The --preload flag tells Gunicorn to "load the app before forking the worker processes". By doing so, each worker is "given a copy of the app, already instantiated by the Master, rather than instantiating the app itself". This means the following code only executes once in the Master process:
rerun_monitor = Scheduler()
rerun_monitor.start()
rerun_monitor.add_interval_job(job_to_be_run,\
seconds=JOB_INTERVAL)
Additionally, we need to set the jobstore to be anything other than :memory:.This way, although each worker is its own independent process unable of communicating with the other 7, by using a local database (rather then memory) we guarantee a single-point-of-truth for CRUD operations on the jobstore.
from apscheduler.schedulers.background import BackgroundScheduler
from apscheduler.jobstores.sqlalchemy import SQLAlchemyJobStore
rerun_monitor = Scheduler(
jobstores={'default': SQLAlchemyJobStore(url='sqlite:///jobs.sqlite')})
rerun_monitor.start()
rerun_monitor.add_interval_job(job_to_be_run,\
seconds=JOB_INTERVAL)
Lastly, we want to use the BackgroundScheduler because of its implementation of start(). When we call start() in the BackgroundScheduler, a new thread is spun up in the background, which is responsible for scheduling/executing jobs. This is significant because remember in step (1), due to our --preload flag we only execute the start() function once, in the Master Gunicorn process. By definition, forked processes do not inherit the threads of their Parent, so each worker doesn't run the BackgroundScheduler thread.
from apscheduler.jobstores.sqlalchemy import SQLAlchemyJobStore
rerun_monitor = BackgroundScheduler(
jobstores={'default': SQLAlchemyJobStore(url='sqlite:///jobs.sqlite')})
rerun_monitor.start()
rerun_monitor.add_interval_job(job_to_be_run,\
seconds=JOB_INTERVAL)
As a result of all of this, every Gunicorn worker has an APScheduler that has been tricked into a "STARTED" state, but actually isn't running because it drops the threads of it's parent! Each instance is also capable of updating the jobstore database, just not executing any jobs!
Check out flask-APScheduler for a quick way to run APScheduler in a web-server (like Gunicorn), and enable CRUD operations for each job.
I found a fix that worked with a Django project having a very similar issue. I simply bind a TCP socket the first time the scheduler starts and check against it subsequently. I think the following code can work for you as well with minor tweaks.
import sys, socket
try:
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.bind(("127.0.0.1", 47200))
except socket.error:
print "!!!scheduler already started, DO NOTHING"
else:
from apscheduler.schedulers.background import BackgroundScheduler
scheduler = BackgroundScheduler()
scheduler.start()
print "scheduler started"
Short answer: You can't do it properly without consequences.
I'm using Gunicorn as an example, but it is essentially the same for uWSGI. There are various hacks when running multiple processes, to name a few:
use --preload option
use on_starting hook to start the APScheduler background scheduler
use when_ready hook to start the APScheduler background scheduler
They work to some extent but may get the following errors:
worker timing out frequently
scheduler hanging when there are no jobs https://github.com/agronholm/apscheduler/issues/305
APScheduler is designed to run in a single process where it has complete control over the process of adding jobs to job stores. It uses threading.Event's wait() and set() methods to coordinate. If they are run by different processes, the coordination wouldn't work.
It is possible to run it in Gunicorn in a single process.
use only one worker process
use the post_worker_init hook to start the scheduler, this will make sure the scheduler is run only in the worker process but not the master process
The author also pointed out sharing the job store amount multiple processes isn't possible. https://apscheduler.readthedocs.io/en/stable/faq.html#how-do-i-share-a-single-job-store-among-one-or-more-worker-processes He also provided a solution using RPyC.
While it's entirely doable to wrap APScheduler with a REST interface. You might want to consider serving it as a standalone app with one worker. In another word, if you have others endpoints, put them in another app where you can use multiple workers.
I've noticed the following in the docs:
Ideally task functions should be idempotent: meaning the function won’t cause unintended effects even if called multiple times with the same arguments. Since the worker cannot detect if your tasks are idempotent, the default behavior is to acknowledge the message in advance, just before it’s executed, so that a task invocation that already started is never executed again.
If your task is idempotent you can set the acks_late option to have the worker acknowledge the message after the task returns instead. See also the FAQ entry Should I use retry or acks_late?.
If set to True messages for this task will be acknowledged after the task has been executed, not just before (the default behavior).
Note: This means the task may be executed multiple times should the worker crash in the middle of execution. Make sure your tasks are idempotent.
Then there is the BROKER_TRANSPORT_OPTIONS = {'confirm_publish': True} option found here. I could not find official documentation for that.
I want to be certain that tasks which are submitted to celery (1) arrive at celery and (2) eventually get executed.
Here is how I think it works:
Celery stores the information which tasks should get executed in a broker (typically RabbitMQ or Redis)
The application (e.g. Django) submits a task to Celery which immediately stores it in the broker. confirm_publish confirms that it was added (right?). If confirm_publish is set but the confirmation is missing, it retries (right?).
Celery takes messages from the broker. Now celery behaves as a consumer for the broker. The consumer acknowledges (confirms) that it received a message an the broker stores this information. If the consumer didn't sent an acknowledgement, the broker will re-try to send it.
Is that correct?
I am a beginner with django, I have celery installed.
I am confused about the working of the celery, if the queued works are handled synchronously or asynchronously. Can other works be queued when the queued work is already being processed?
Celery is a task queuing system, that is backed by a message queuing system, Celery allows you to invoke tasks asynchronously, in a way that won't block your process for the task to finish, you can wait for the task to finish using the AsyncResult.get.
Other tasks can be queued while a task is being processed, and if Celery is running more than one process/thread (which is the default case), tasks will be executed in parallel to each others.
It is your responsibility to make sure that related tasks are executed in the correct order, e.g. if the output of a task A is an input to the other task B then you should make sure that you get the result from task A before you start the task B.
Read Avoid launching synchronous subtasks from Celery documentation.
I think you're possibly a bit confused about what Celery does.
Celery isn't really responsible for queueing at all. That is taken care of by the queue itself - RabbitMQ, Redis, or whatever. The only way Celery gets involved at this end is as a library that you call inside your app to serialize to task into something suitable for putting onto the queue. Since that is done by your web app, it is exactly as synchronous or asynchronous as your app itself: usually, in production, you'd have multiple processes running your site, each of those could put things onto the queue simultaneously, but each queueing action is done in-process.
The main point of Celery is the separate worker processes. This is where the asynchronous bit comes from: the workers run completely separately from your web app, and pick tasks off the queue as necessary. They are not at all involved in the process of putting tasks onto the queue in the first place.
1) I am currently working on a web application that exposes a REST api and uses Django and Celery to handle request and solve them. For a request in order to get solved, there have to be submitted a set of celery tasks to an amqp queue, so that they get executed on workers (situated on other machines). Each task is very CPU intensive and takes very long (hours) to finish.
I have configured Celery to use also amqp as results-backend, and I am using RabbitMQ as Celery's broker.
Each task returns a result that needs to be stored afterwards in a DB, but not by the workers directly. Only the "central node" - the machine running django-celery and publishing tasks in the RabbitMQ queue - has access to this storage DB, so the results from the workers have to return somehow on this machine.
The question is how can I process the results of the tasks execution afterwards? So after a worker finishes, the result from it gets stored in the configured results-backend (amqp), but now I don't know what would be the best way to get the results from there and process them.
All I could find in the documentation is that you can either check on the results's status from time to time with:
result.state
which means that basically I need a dedicated piece of code that runs periodically this command, and therefore keeps busy a whole thread/process only with this, or to block everything with:
result.get()
until a task finishes, which is not what I wish.
The only solution I can think of is to have on the "central node" an extra thread that runs periodically a function that basically checks on the async_results returned by each task at its submission, and to take action if the task has a finished status.
Does anyone have any other suggestion?
Also, since the backend-results' processing takes place on the "central node", what I aim is to minimize the impact of this operation on this machine.
What would be the best way to do that?
2) How do people usually solve the problem of dealing with the results returned from the workers and put in the backend-results? (assuming that a backend-results has been configured)
I'm not sure if I fully understand your question, but take into account each task has a task id. If tasks are being sent by users you can store the ids and then check for the results using json as follows:
#urls.py
from djcelery.views import is_task_successful
urlpatterns += patterns('',
url(r'(?P<task_id>[\w\d\-\.]+)/done/?$', is_task_successful,
name='celery-is_task_successful'),
)
Other related concept is that of signals each finished task emits a signal. A finnished task will emit a task_success signal. More can be found on real time proc.
I have a long running task which is NOT async, it will block the response of django, the server stack is nginx+uwsgi, what happens after nginx decide it's timeout, will my task (the uwsgi worker and the django view thread) be killed?
normally the request will continue in uWSGI until a "bad condition" (like sending a chunk of the response to a disconnected client) will fail. It is a configurable behaviour, but i suggest you to not touch it if you have no VERY specific reasons