feat: improve celery stability

app/views/views.py

@@ -2,8 +2,8 @@
 import logging
 
 from celery.exceptions import TimeoutError
-from celery.result import AsyncResult
-from celery.states import FAILURE, PENDING, SUCCESS
+# from celery.result import AsyncResult
+from celery.states import FAILURE, PENDING, STARTED, SUCCESS
 from django.contrib.auth.decorators import login_required
 from django.core import serializers
 from django.http import (
@@ -26,6 +26,7 @@
 from app.worker.tasks import receiptor
 from app.worker.tasks.exporter import exporter
 from app.worker.tasks.importers import historical_data_importer
+from reboot.celery import app
 
 logger = logging.getLogger(__name__)
 
@@ -121,11 +122,12 @@ def poll_state(request: HttpRequest):
             request=request,
             err_msg="The task_id query parameter of the request was omitted.")
 
-    task = AsyncResult(task_id)
+    task = app.AsyncResult(task_id)
     res = JsonResponse(_poll_state(PENDING, 0, 200))
+    print(f"!!! task id={task_id},state={task.state},successful={task.successful()},ready={task.ready()},failed={task.failed()}")
     if task.state == FAILURE or task.failed():
         res = JsonResponse(_poll_state(FAILURE, 0, 400))
-    elif task.state == PROGRESS:
+    elif task.state == STARTED:
         res = JsonResponse(task.result) if isinstance(
             task.result, dict) else HttpResponse(task.result)
     elif task.state == SUCCESS or task.successful() or task.ready():
@@ -146,8 +148,9 @@ def download_file(request: HttpRequest):
         while (attempts < ATTEMPT_LIMIT):
             try:
                 attempts += 1
-                task = AsyncResult(task_id)
-                result = task.get(timeout=0.5 * attempts)
+                task = app.AsyncResult(task_id)
+                print(f"!!! task id={task_id},state={task.state},successful={task.successful()},ready={task.ready()},failed={task.failed()}")
+                result = task.get(timeout=1.0 * attempts)
                 print(f"{task} {task_name} success #{attempts}: {result}")
                 break
             except TimeoutError:

app/worker/app_celery.py

@@ -1,33 +1,33 @@
 import traceback
-import celery
-from celery.states import SUCCESS, FAILURE
 from http import HTTPStatus
 
-PROGRESS = 'PROGRESS'
+from celery.states import FAILURE, STARTED, SUCCESS
+
+from reboot.celery import app
 
 ATTEMPT_LIMIT = 5
 
 
 def update_state(state, percent, http_status):
     print('{0!r} state: {1!r}, progress: {2!r}'.format(
-        celery.current_task.request.id, state, percent))
-    celery.current_task.update_state(state=state, meta={
+        app.current_task.request.id, state, percent))
+    app.current_task.update_state(state=state, meta={
         'state': state,
         'process_percent': percent,
         'status': http_status,
     })
 
 
 def update_percent(percent):
-    update_state(PROGRESS, percent, HTTPStatus.ACCEPTED)
+    update_state(STARTED, percent, HTTPStatus.ACCEPTED)
 
 
 def set_success():
     update_state(SUCCESS, 100, HTTPStatus.OK)
 
 
 def set_failure(e):
-    celery.current_task.update_state(
+    app.current_task.update_state(
         state=FAILURE,
         meta={
             'exc_type': type(e).__name__,
@@ -38,7 +38,7 @@ def set_failure(e):
         })
 
 
-class AppTask(celery.Task):
+class AppTask(app.Task):
     max_retries = 0
     # default_retry_delay = 10
 

app/worker/tasks/__init__.py

@@ -1,13 +1,14 @@
 '''
 Module for tasks to be sent on task queue
 '''
-from celery import task
-
 from app.worker.app_celery import AppTask
+# from celery import task
+from reboot.celery import app
+
 from .create_receipt import Receiptor
 
 
-@task(bind=True, base=AppTask)
+@app.task(bind=True, base=AppTask)
 def receiptor(self, queryset, total_count):
     receiptor = Receiptor(queryset, total_count)
     return receiptor()

app/worker/tasks/exporter.py

@@ -1,16 +1,17 @@
 import csv
 
-from celery import task
+# from celery import task
 from celery.utils.log import get_task_logger
 from django.core import serializers
 from django.db.models.query import QuerySet
 from django.http import HttpResponse
 
 from app.constants.field_names import CURRENT_FIELDS
 from app.worker.app_celery import AppTask, update_percent
+from reboot.celery import app
 
 
-@task(bind=True, base=AppTask)
+@app.task(bind=True, base=AppTask)
 def exporter(self, file_name, qs: QuerySet = None, total_count: int = 0):
     rows = serializers.deserialize('json', qs)
     csv_exporter = CsvExporter(file_name, rows, total_count)

app/worker/tasks/importers/__init__.py

@@ -1,13 +1,15 @@
 """
 Module for csv file importers to be sent to queue
 """
-from celery import task
+# from celery import task
 
 from app.worker.app_celery import AppTask
+from reboot.celery import app
+
 from .historical_data_importer import HistoricalDataImporter
 
 
-@task(bind=True, base=AppTask)
+@app.task(bind=True, base=AppTask)
 def historical_data_importer(self, csvpath):
     importer = HistoricalDataImporter(csvpath)
     importer()

reboot/celeryconfig.py

@@ -8,11 +8,13 @@
 broker_pool_limit = 1
 event_queue_expires = 60
 worker_prefetch_multiplier = 1
-worker_concurrency = 10
+worker_concurrency = 1
 accept_content = ['json', 'pickle']
 result_backend = config("REDIS_URL")
 task_serializer = 'pickle'
 result_serializer = 'pickle'
+task_track_started = True
+task_ignore_result = False
 
 # Use PROD settings if valid CLOUDAMQP_URl, else dev
 if config('CLOUDAMQP_URL', default=False):