Question

我正在尝试使用Locust对使用Clarifai API创建的模型进行压力测试。我似乎大部分工作都在工作，但似乎无法使任务分配的语法正常工作。 UI在起作用，蝗虫群在起作用，我可以在csv中获取统计信息，但是它没有执行此处定义的任务，这是对自定义模型的预测。

这是我进行实际通话的功能。

import gevent
from gevent import monkey
monkey.patch_all()

from clarifai.rest import ClarifaiApp, Workflow
import random
from locust import task, TaskSet, between, User

import app_settings. #contains some hardcoded values as a script. 
from locust_utils import ClarifaiLocust, locust_call
      

class ApiUser(ClarifaiLocust):
    min_wait = 1000
    max_wait = 3000
    wait_time = between(2,5)

    def on_start(self):
        # self.small_model = self.client.get_app('model-specialization-demo-pt2').workflows.get('locust-vehicle-det')
        self.small_model = self.client.get_app('app_id').workflows.get('locust-vehicle-det')

    @task
    def predict_small_model(self):
        locust_call(
            self.small_model.predict_by_url,
            'predict to Public Vehicle Detector',
            url=random.choice(app_settings.small_app_urls)

        )

下面称为ClarifaiLocust和locust_call的函数

def locust_call(func, name, *args, **kwargs):
  start_time = time.time()
  try:
    func(*args, **kwargs)  # Don't really care about results for stress test
  except ApiError as e:
    total_time = int((time.time() - start_time) * 1000)
    events.request_failure.fire(
        request_type='Client', name=name, response_time=total_time, exception=e)
  else:
    total_time = int((time.time() - start_time) * 1000)
    events.request_success.fire(
        request_type='Client', name=name, response_time=total_time, response_length=0)


class ClarifaiLocust(HttpUser):
  test_app = None
  search_app = None
  wait_time = between(2,5)

  def __init__(self, *args, **kwargs):
    # Locust.__init__(self, *args, **kwargs)
    User.__init__(self, *args, **kwargs)
    #super(ClarifaiLocust, self).__init__(*args, **kwargs)
    self.client = ClarifaiUser(self.host)

但我一直收到此错误。

raise Exception("No tasks defined. use the @task decorator or set the tasks property of the User")
Exception: No tasks defined. use the @task decorator or set the tasks property of the User

我在做什么错了？

Answer 1

在ClarifaiLocust上添加abstract = True，否则Locust也会尝试运行它。

所以：

class ClarifaiLocust(HttpUser):
    abstract = True
    ....

似乎无法使用Clarifai API为蝗虫设置任务

1 个答案: