ext_celery.py 3.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293
  1. from datetime import timedelta
  2. import pytz
  3. from celery import Celery, Task
  4. from celery.schedules import crontab
  5. from flask import Flask
  6. from configs import dify_config
  7. def init_app(app: Flask) -> Celery:
  8. class FlaskTask(Task):
  9. def __call__(self, *args: object, **kwargs: object) -> object:
  10. with app.app_context():
  11. return self.run(*args, **kwargs)
  12. broker_transport_options = {}
  13. if dify_config.CELERY_USE_SENTINEL:
  14. broker_transport_options = {
  15. "master_name": dify_config.CELERY_SENTINEL_MASTER_NAME,
  16. "sentinel_kwargs": {
  17. "socket_timeout": dify_config.CELERY_SENTINEL_SOCKET_TIMEOUT,
  18. },
  19. }
  20. celery_app = Celery(
  21. app.name,
  22. task_cls=FlaskTask,
  23. broker=dify_config.CELERY_BROKER_URL,
  24. backend=dify_config.CELERY_BACKEND,
  25. task_ignore_result=True,
  26. )
  27. # Add SSL options to the Celery configuration
  28. ssl_options = {
  29. "ssl_cert_reqs": None,
  30. "ssl_ca_certs": None,
  31. "ssl_certfile": None,
  32. "ssl_keyfile": None,
  33. }
  34. celery_app.conf.update(
  35. result_backend=dify_config.CELERY_RESULT_BACKEND,
  36. broker_transport_options=broker_transport_options,
  37. broker_connection_retry_on_startup=True,
  38. worker_log_format=dify_config.LOG_FORMAT,
  39. worker_task_log_format=dify_config.LOG_FORMAT,
  40. worker_hijack_root_logger=False,
  41. timezone=pytz.timezone(dify_config.LOG_TZ),
  42. )
  43. if dify_config.BROKER_USE_SSL:
  44. celery_app.conf.update(
  45. broker_use_ssl=ssl_options, # Add the SSL options to the broker configuration
  46. )
  47. if dify_config.LOG_FILE:
  48. celery_app.conf.update(
  49. worker_logfile=dify_config.LOG_FILE,
  50. )
  51. celery_app.set_default()
  52. app.extensions["celery"] = celery_app
  53. imports = [
  54. "schedule.clean_embedding_cache_task",
  55. "schedule.clean_unused_datasets_task",
  56. "schedule.create_tidb_serverless_task",
  57. "schedule.update_tidb_serverless_status_task",
  58. ]
  59. day = dify_config.CELERY_BEAT_SCHEDULER_TIME
  60. beat_schedule = {
  61. "clean_embedding_cache_task": {
  62. "task": "schedule.clean_embedding_cache_task.clean_embedding_cache_task",
  63. "schedule": timedelta(days=day),
  64. },
  65. "clean_unused_datasets_task": {
  66. "task": "schedule.clean_unused_datasets_task.clean_unused_datasets_task",
  67. "schedule": timedelta(days=day),
  68. },
  69. "create_tidb_serverless_task": {
  70. "task": "schedule.create_tidb_serverless_task.create_tidb_serverless_task",
  71. "schedule": crontab(minute="0", hour="*"),
  72. },
  73. "update_tidb_serverless_status_task": {
  74. "task": "schedule.update_tidb_serverless_status_task.update_tidb_serverless_status_task",
  75. "schedule": crontab(minute="30", hour="*"),
  76. },
  77. }
  78. celery_app.conf.update(beat_schedule=beat_schedule, imports=imports)
  79. return celery_app