Airflow速用

Airflow是Apache用python编写的，用到了 flask框架及相关插件,rabbitmq,celery等（windows不兼容）；、

 

主要实现的功能

• 编写 定时任务，及任务间的编排；
• 提供了web界面 可以手动触发任务，分析任务执行顺序，任务执行状态，任务代码，任务日志等等；
• 实现celery的分布式任务调度系统;
• 简单方便的实现了 任务在各种状态下触发 发送邮件的功能；https://airflow.apache.org/concepts.html#email-configuration
• 对组合任务 可以根据 不同参数进入不同分支进行处理 http://airflow.apache.org/concepts.html#branching
• 执行 bash脚本命令；
• 对组合任务 设置触发条件（如:全部失败/成功时执行某任务 等等）http://airflow.apache.org/concepts.html#trigger-rules
• 简单实现随机 负载均衡和容错能力 http://airflow.apache.org/concepts.html#connections
• 对组合任务 间进行数据传递 http://airflow.apache.org/concepts.html#xcoms
• 对分布式任务指定 queue, worker可以指定消费的queue(celery的使用) http://airflow.apache.org/concepts.html#queues
• 存储日志到远程 http://airflow.apache.org/howto/write-logs.html
• 调用 远程 谷歌云，亚马逊云 相关服务（如语音识别等等）https://airflow.apache.org/integration.html#integration
• 调用 钉钉 相关服务

实现功能总结

不仅celery有的功能我都有， 我还能通过页面手动触发/暂停任务，管理任务特方便；我他妈还能 调用谷歌云等服务，日志也能方便打印到云服务上。。。。。。；我就是牛！

 

核心思想

• DAG:英文为：Directed Acyclic Graph；指 (有向无环图)有向非循环图，是想运行的一系列任务的集合，不关心任务是做什么的，只关心 任务间的组成方式，确保在正确的时间，正确的顺序触发各个任务，准确的处理意外情况；http://airflow.apache.org/concepts.html#dags
• DAGs:多个任务集（多个DAG）
• Operator: 指 某些类型任务的模板 类；如 PythonOperator(执行python相关操作)，EmailOperator(执行发送邮件相关操作)，SimpleHttpOperator(执行发送http请求相关操作) 等几十种（源码可见）http://airflow.apache.org/howto/operator/index.html#
• Task:当通过 Operator定义了执行任务内容后，在实例化后，便是 Task,为DAG中任务集合的具体任务
• Executor：数据库记录任务状态(排队queued，预执行scheduled，运行中running，成功success，失败failed),调度器（Scheduler ）从数据库取数据并决定哪些需要完成，然后 Executor 和调度器一起合作，给任务需要的资源让其完成。Executor间(如 LocalExecutor,CeleryExecutor)不同点在于他们拥有不同的资源以及如何利用资源分配工作，如LocalExecutor只在本地并行执行任务，CeleryExecutor分布式多机器执行任务。 https://www.astronomer.io/guides/airflow-executors-explained/
• Hook：是airflow与外部平台/数据库交互的方式，如 http/ssh/sftp等等，是Operator的基础部分(如SimpleHttpOperator 需要依赖HttpHook)

 

任务间定义排序的方法

官方推荐使用 移位操作符 方法，因为较为直观，容易理解

如：  op1 >> op2 >> op3   表示任务执行顺序为  从左到右依次执行

官方文档介绍：http://airflow.apache.org/concepts.html#bitshift-composition

 

提高airflow相关执行速度方法

通过修改airflow.cfg相关配置

官方文档如下：http://airflow.apache.org/faq.html

 

安装及启动相关服务

• 创建python虚拟环境 venv
• 添加airflow.cfg（此配置注解在下面）的配置文件夹路径：先 vi venv/bin/active; 里面输入 export AIRFLOW_HOME="/mnt/e/project/airflow_config/local"

• 命令行：pip install apache-airflow

• 根据airflow.cfg的数据库配置，在连接的数据库服务创建一个 名为 airflow_db的数据库

• 命令行初始化数据库：airflow initdb

• 命令行启动web服务: airflow webserver -p 8080

• 命令行启动任务调度服务：airflow scheduler

• 命令行启动worker：airflow worker -q queue_name

 

使用 http_operator发送http请求并在失败时,发送邮件

1.设置邮件html模板（如下为自定义模板）

<h2 style="color: red">Xxx service task exception,please fix them!!!h2>
Try {{try_number}} out of {{max_tries + 1}}<br><br>
<b>dag id: b>{{ti.dag_id}}<br><br>
<b>task id: b>{{ti.task_id}}<br><br>
<b>task state: b>{{ti.state}}<br><br>

<b>Exception:b>
<p style="color: #0d7bdc">{{exception_html}}p>
<b>Log Url: b>
<a href="{{ti.log_url}}" style="color: red">Linka><br><br>
<b>Host: b>
{{ti.hostname}}<br><br>
<b>Log file path: b> {{ti.log_filepath}}<br><br>
<b>Mark success: b> <a href="{{ti.mark_success_url}}">Linka><br>

模板效果图：

 2. airflow.cfg文件中配置 发送邮件服务

 

 3.编写代码：

 1 # -*- coding: utf-8 -*-
 2 """
 3 (C) xxx 
 4 All rights reserved
 5 create time '2019/10/21 09:27'
 6 """
 7 import os
 8 from datetime import datetime
 9 
10 import pytz
11 from airflow import DAG
12 from airflow.models import Variable
13 from airflow.operators.http_operator import SimpleHttpOperator
14 
15 # 设置第一次触发任务时间 及 设置任务执行的时区
16 tz = pytz.timezone("Asia/Shanghai")
17 dt = datetime(2019, 10, 11, 0, 0, tzinfo=tz)
18 utc_dt = dt.astimezone(pytz.utc).replace(tzinfo=None)
19 
20 # 从环境变量找到 当前环境
21 env = os.environ.get("PROJECT_ENV", "LOCAL")
22 # 添加 需要的相关环境变量，可在 web网页中设置；注意 变量名 以AIRFLOW_CONN_开头，并且大写
23 os.environ["AIRFLOW_CONN_OLY_HOST"] = Variable.get("OLY_HOST_%s" % env)
24 
25 # dag默认参数
26 args = {
27     "owner": "Rgc",  # 任务拥有人
28     "depends_on_past": False,  # 是否依赖过去执行此任务的结果，如果为True,则过去任务必须成功，才能执行此次任务
29     "start_date": utc_dt,  # 任务开始执行时间
30     "email": ["[email protected]"],  # 邮件地址，可以填写多个
31     "email_on_failure": True,  # 触发邮件发送的 时机，此处为失败时触发
32 }
33 
34 # 定义一个DAG
35 # 参数catchup指 是否填充执行 start_date到现在 未执行的缺少任务；如：start_date定义为2019-10-10,现在是2019-10-29,任务是每天定时执行一次，
36 # 如果此参数设置为True，则 会生成 10号到29号之间的19此任务；如果设置为False,则不会补充执行任务；
37 # schedule_interval：定时执行方式，推荐使用如下字符串方式, 方便写出定时规则的网址：https://crontab.guru/
38 dag = DAG("HttpSendDag", catchup=False, default_args=args, schedule_interval="0 19 * * *")
39 # 设置 dag文档注释，可在web界面任务详情中看到
40 dag.doc_md = __doc__
41 
42 # 定义此 http operator相关详情，详细使用方法 可访问此类定义__init__()方法
43 task = SimpleHttpOperator(
44     task_id="task_http_send",  # 任务id
45     http_conn_id="oly_host",  # http请求地址，值为上面23行定义
46     method="POST",  # http请求方法
47     endpoint="user/manage",  # http请求路径
48     dag=dag  # 任务所属dag
49 )
50 # 定义任务 文档注释，可在web界面任务详情中看到
51 task.doc_md = f"""\
52 #Usage
53 此任务主要向Project服务({Variable.get("OLY_HOST_%s" % env)})发送http请求，每天晚上7点定时运行!
54 """

 

任务间数据交流方法

    使用Xcoms(cross-communication),类似于redis存储结构,任务推送数据或者从中下拉数据，数据在任务间共享

    推送数据主要有2中方式：1：使用xcom_push()方法  2：直接在PythonOperator中调用的函数 return即可

    下拉数据 主要使用 xcom_pull()方法

 官方代码示例及注释：

 1 from __future__ import print_function
 2 
 3 import airflow
 4 from airflow import DAG
 5 from airflow.operators.python_operator import PythonOperator
 6 
 7 args = {
 8     'owner': 'airflow',
 9     'start_date': airflow.utils.dates.days_ago(2),
10     'provide_context': True,
11 }
12 
13 dag = DAG('example_xcom', schedule_interval="@once", default_args=args)
14 
15 value_1 = [1, 2, 3]
16 value_2 = {'a': 'b'}
17 
18 
19 # 2种推送数据的方式，分别为xcom_push，和直接return
20 
21 def push(**kwargs):
22     """Pushes an XCom without a specific target"""
23     kwargs['ti'].xcom_push(key='value from pusher 1', value=value_1)
24 
25 
26 def push_by_returning(**kwargs):
27     """Pushes an XCom without a specific target, just by returning it"""
28     return value_2
29 
30 
31 def puller(**kwargs):
32     """
33     下拉数据的方法
34     :param kwargs:
35     :return:
36     """
37     ti = kwargs['ti']
38 
39     # get value_1
40     v1 = ti.xcom_pull(key=None, task_ids='push')
41     assert v1 == value_1
42 
43     # get value_2
44     v2 = ti.xcom_pull(task_ids='push_by_returning')
45     assert v2 == value_2
46 
47     # get both value_1 and value_2
48     v1, v2 = ti.xcom_pull(key=None, task_ids=['push', 'push_by_returning'])
49     assert (v1, v2) == (value_1, value_2)
50 
51 
52 push1 = PythonOperator(
53     task_id='push',
54     dag=dag,
55     python_callable=push,
56 )
57 
58 push2 = PythonOperator(
59     task_id='push_by_returning',
60     dag=dag,
61     python_callable=push_by_returning,
62 )
63 
64 pull = PythonOperator(
65     task_id='puller',
66     dag=dag,
67     python_callable=puller,
68 )
69 
70 # 任务执行顺序为
71 # push1 >> pull
72 # push2 >> pull
73 
74 pull << [push1, push2]

View Code

 

开启 web网页登录需要用户名密码功能

1.airflow.cfg文件修改

# 设置为True
rbac = True

2.重启airflow相关服务

3.通过 命令行 添加 用户

airflow create_user -r Admin -e [email protected] -f A -l dmin -u admin -p passwd

4.访问页面，输入用户名，密码即可

 

忽略某些DAG文件,不调用

在dag任务文件夹下，添加一个 .airflowignore文件(像 .gitignore），里面写 文件名即可（支持正则）

 

 启动及关闭airflow内置 dag示例方法（能够快速学习Airflow）

 开启:修改airflow.cfg配置文件  load_examples = True  并重启即可

 关闭:修改airflow.cfg配置文件  load_examples = True，并清空数据库，并重启即可

 效果图：

 

 

 

airflow配置文件 相关中文注解:

  1 [core]
  2 # The folder where your airflow pipelines live, most likely a
  3 # subfolder in a code repository
  4 # This path must be absolute
  5 # 绝对路径下 一系列dags存放位置，airflow只会从此路径 文件夹下找dag任务
  6 dags_folder = /mnt/e/airflow_project/dags
  7 
  8 # The folder where airflow should store its log files
  9 # This path must be absolute
 10 # 绝对路径下的日志文件夹位置
 11 base_log_folder = /mnt/e/airflow_project/log/
 12 
 13 # Airflow can store logs remotely in AWS S3, Google Cloud Storage or Elastic Search.
 14 # Users must supply an Airflow connection id that provides access to the storage
 15 # location. If remote_logging is set to true, see UPDATING.md for additional
 16 # configuration requirements.
 17 remote_logging = False
 18 remote_log_conn_id =
 19 remote_base_log_folder =
 20 encrypt_s3_logs = False
 21 
 22 # Logging level
 23 logging_level = INFO
 24 fab_logging_level = WARN
 25 
 26 # Logging class
 27 # Specify the class that will specify the logging configuration
 28 # This class has to be on the python classpath
 29 # logging_config_class = my.path.default_local_settings.LOGGING_CONFIG
 30 logging_config_class =
 31 
 32 # Log format
 33 # Colour the logs when the controlling terminal is a TTY.
 34 colored_console_log = True
 35 colored_log_format = [%%(blue)s%%(asctime)s%%(reset)s] {%%(blue)s%%(filename)s:%%(reset)s%%(lineno)d} %%(log_color)s%%(levelname)s%%(reset)s - %%(log_color)s%%(message)s%%(reset)s
 36 colored_formatter_class = airflow.utils.log.colored_log.CustomTTYColoredFormatter
 37 
 38 log_format = [%%(asctime)s] {%%(filename)s:%%(lineno)d} %%(levelname)s - %%(message)s
 39 simple_log_format = %%(asctime)s %%(levelname)s - %%(message)s
 40 
 41 # Log filename format
 42 # 实际处理任务日志 相关
 43 log_filename_template = {{ ti.dag_id }}/{{ ti.task_id }}/{{ ts }}/{{ try_number }}.log
 44 log_processor_filename_template = {{ filename }}.log
 45 # dag处理日志 绝对路径，精确到日志文件
 46 dag_processor_manager_log_location = /mnt/e/airflow_project/log/dag_processor_manager.log
 47 
 48 # Hostname by providing a path to a callable, which will resolve the hostname
 49 # The format is "package:function". For example,
 50 # default value "socket:getfqdn" means that result from getfqdn() of "socket" package will be used as hostname
 51 # No argument should be required in the function specified.
 52 # If using IP address as hostname is preferred, use value "airflow.utils.net:get_host_ip_address"
 53 hostname_callable = socket:getfqdn
 54 
 55 # Default timezone in case supplied date times are naive
 56 # can be utc (default), system, or any IANA timezone string (e.g. Europe/Amsterdam)
 57 # 默认时区，改为上海，然而 没卵用
 58 default_timezone = Asia/Shanghai
 59 
 60 # The executor class that airflow should use. Choices include
 61 # SequentialExecutor, LocalExecutor, CeleryExecutor, DaskExecutor, KubernetesExecutor
 62 # 指定executor(任务分配执行方式)
 63 executor = CeleryExecutor
 64 
 65 # The SqlAlchemy connection string to the metadata database.
 66 # SqlAlchemy supports many different database engine, more information
 67 # their website
 68 # 存储airflow相关数据的 数据库路径
 69 sql_alchemy_conn = mysql+pymysql://root:[email protected]:3306/airflow_db
 70 
 71 # The encoding for the databases
 72 sql_engine_encoding = utf-8
 73 
 74 # If SqlAlchemy should pool database connections.
 75 sql_alchemy_pool_enabled = True
 76 
 77 # The SqlAlchemy pool size is the maximum number of database connections
 78 # in the pool. 0 indicates no limit.
 79 sql_alchemy_pool_size = 5
 80 
 81 # The maximum overflow size of the pool.
 82 # When the number of checked-out connections reaches the size set in pool_size,
 83 # additional connections will be returned up to this limit.
 84 # When those additional connections are returned to the pool, they are disconnected and discarded.
 85 # It follows then that the total number of simultaneous connections the pool will allow is pool_size + max_overflow,
 86 # and the total number of "sleeping" connections the pool will allow is pool_size.
 87 # max_overflow can be set to -1 to indicate no overflow limit;
 88 # no limit will be placed on the total number of concurrent connections. Defaults to 10.
 89 sql_alchemy_max_overflow = 10
 90 
 91 # The SqlAlchemy pool recycle is the number of seconds a connection
 92 # can be idle in the pool before it is invalidated. This config does
 93 # not apply to sqlite. If the number of DB connections is ever exceeded,
 94 # a lower config value will allow the system to recover faster.
 95 sql_alchemy_pool_recycle = 1800
 96 
 97 # How many seconds to retry re-establishing a DB connection after
 98 # disconnects. Setting this to 0 disables retries.
 99 sql_alchemy_reconnect_timeout = 300
100 
101 # The schema to use for the metadata database
102 # SqlAlchemy supports databases with the concept of multiple schemas.
103 sql_alchemy_schema =
104 
105 # The amount of parallelism as a setting to the executor. This defines
106 # the max number of task instances that should run simultaneously
107 # on this airflow installation
108 parallelism = 32
109 
110 # The number of task instances allowed to run concurrently by the scheduler
111 dag_concurrency = 16
112 
113 # Are DAGs paused by default at creation
114 dags_are_paused_at_creation = True
115 
116 # The maximum number of active DAG runs per DAG
117 max_active_runs_per_dag = 16
118 
119 # Whether to load the examples that ship with Airflow. It's good to
120 # get started, but you probably want to set this to False in a production
121 # environment
122 load_examples = False
123 
124 # Where your Airflow plugins are stored
125 # 自定义 界面及api所在 绝对路径文件夹 官网用法： http://airflow.apache.org/plugins.html
126 plugins_folder = /mnt/e/airflow_project/plugins
127 
128 # Secret key to save connection passwords in the db
129 # 对使用到的 连接密码 进行加密，此为秘钥 官网用法： https://airflow.apache.org/howto/secure-connections.html
130 fernet_key = Et8ULvn0biL8X0xXl66wHawhdetf7utIDYDgNzZh4nCnE=
131 
132 # Whether to disable pickling dags
133 donot_pickle = False
134 
135 # How long before timing out a python file import while filling the DagBag
136 dagbag_import_timeout = 30
137 
138 # The class to use for running task instances in a subprocess
139 task_runner = StandardTaskRunner
140 
141 # If set, tasks without a `run_as_user` argument will be run with this user
142 # Can be used to de-elevate a sudo user running Airflow when executing tasks
143 default_impersonation =
144 
145 # What security module to use (for example kerberos):
146 security =
147 
148 # If set to False enables some unsecure features like Charts and Ad Hoc Queries.
149 # In 2.0 will default to True.
150 secure_mode = False
151 
152 # Turn unit test mode on (overwrites many configuration options with test
153 # values at runtime)
154 unit_test_mode = False
155 
156 # Name of handler to read task instance logs.
157 # Default to use task handler.
158 task_log_reader = task
159 
160 # Whether to enable pickling for xcom (note that this is insecure and allows for
161 # RCE exploits). This will be deprecated in Airflow 2.0 (be forced to False).
162 enable_xcom_pickling = True
163 
164 # When a task is killed forcefully, this is the amount of time in seconds that
165 # it has to cleanup after it is sent a SIGTERM, before it is SIGKILLED
166 killed_task_cleanup_time = 60
167 
168 # Whether to override params with dag_run.conf. If you pass some key-value pairs through `airflow backfill -c` or
169 # `airflow trigger_dag -c`, the key-value pairs will override the existing ones in params.
170 dag_run_conf_overrides_params = False
171 
172 # Worker initialisation check to validate Metadata Database connection
173 worker_precheck = False
174 
175 # When discovering DAGs, ignore any files that don't contain the strings `DAG` and `airflow`.
176 dag_discovery_safe_mode = True
177 
178 
179 [cli]
180 # In what way should the cli access the API. The LocalClient will use the
181 # database directly, while the json_client will use the api running on the
182 # webserver
183 api_client = airflow.api.client.local_client
184 
185 # If you set web_server_url_prefix, do NOT forget to append it here, ex:
186 # endpoint_url = http://localhost:8080/myroot
187 # So api will look like: http://localhost:8080/myroot/api/experimental/...
188 endpoint_url = http://localhost:18080
189 
190 [api]
191 # How to authenticate users of the API
192 auth_backend = airflow.api.auth.backend.default
193 
194 [lineage]
195 # what lineage backend to use
196 backend =
197 
198 [atlas]
199 sasl_enabled = False
200 host =
201 port = 21000
202 username =
203 password =
204 
205 [operators]
206 # The default owner assigned to each new operator, unless
207 # provided explicitly or passed via `default_args`
208 default_owner = airflow
209 default_cpus = 1
210 default_ram = 512
211 default_disk = 512
212 default_gpus = 0
213 
214 [hive]
215 # Default mapreduce queue for HiveOperator tasks
216 default_hive_mapred_queue =
217 
218 [webserver]
219 # web端访问配置
220 # The base url of your website as airflow cannot guess what domain or
221 # cname you are using. This is used in automated emails that
222 # airflow sends to point links to the right web server
223 base_url = http://localhost:18080
224 
225 # The ip specified when starting the web server
226 web_server_host = 0.0.0.0
227 
228 # The port on which to run the web server
229 web_server_port = 18080
230 
231 # Paths to the SSL certificate and key for the web server. When both are
232 # provided SSL will be enabled. This does not change the web server port.
233 web_server_ssl_cert =
234 web_server_ssl_key =
235 
236 # Number of seconds the webserver waits before killing gunicorn master that doesn't respond
237 web_server_master_timeout = 120
238 
239 # Number of seconds the gunicorn webserver waits before timing out on a worker
240 web_server_worker_timeout = 120
241 
242 # Number of workers to refresh at a time. When set to 0, worker refresh is
243 # disabled. When nonzero, airflow periodically refreshes webserver workers by
244 # bringing up new ones and killing old ones.
245 worker_refresh_batch_size = 1
246 
247 # Number of seconds to wait before refreshing a batch of workers.
248 worker_refresh_interval = 30
249 
250 # Secret key used to run your flask app
251 secret_key = temporary_key
252 
253 # Number of workers to run the Gunicorn web server
254 workers = 4
255 
256 # The worker class gunicorn should use. Choices include
257 # sync (default), eventlet, gevent
258 worker_class = sync
259 
260 # Log files for the gunicorn webserver. '-' means log to stderr.
261 access_logfile = -
262 error_logfile = -
263 
264 # Expose the configuration file in the web server
265 # This is only applicable for the flask-admin based web UI (non FAB-based).
266 # In the FAB-based web UI with RBAC feature,
267 # access to configuration is controlled by role permissions.
268 expose_config = False
269 
270 # Set to true to turn on authentication:
271 # https://airflow.apache.org/security.html#web-authentication
272 authenticate = False
273 
274 # Filter the list of dags by owner name (requires authentication to be enabled)
275 filter_by_owner = False
276 
277 # Filtering mode. Choices include user (default) and ldapgroup.
278 # Ldap group filtering requires using the ldap backend
279 #
280 # Note that the ldap server needs the "memberOf" overlay to be set up
281 # in order to user the ldapgroup mode.
282 owner_mode = user
283 
284 # Default DAG view.  Valid values are:
285 # tree, graph, duration, gantt, landing_times
286 dag_default_view = tree
287 
288 # Default DAG orientation. Valid values are:
289 # LR (Left->Right), TB (Top->Bottom), RL (Right->Left), BT (Bottom->Top)
290 dag_orientation = LR
291 
292 # Puts the webserver in demonstration mode; blurs the names of Operators for
293 # privacy.
294 demo_mode = False
295 
296 # The amount of time (in secs) webserver will wait for initial handshake
297 # while fetching logs from other worker machine
298 log_fetch_timeout_sec = 5
299 
300 # By default, the webserver shows paused DAGs. Flip this to hide paused
301 # DAGs by default
302 hide_paused_dags_by_default = False
303 
304 # Consistent page size across all listing views in the UI
305 page_size = 100
306 
307 # Use FAB-based webserver with RBAC feature
308 # 是否登录时 需要用户名 密码 验证功能；https://airflow.apache.org/security.html#rbac-ui-security
309 rbac = False
310 
311 # Define the color of navigation bar
312 navbar_color = #007A87
313 
314 # Default dagrun to show in UI
315 default_dag_run_display_number = 25
316 
317 # Enable werkzeug `ProxyFix` middleware
318 enable_proxy_fix = False
319 
320 # Set secure flag on session cookie
321 cookie_secure = False
322 
323 # Set samesite policy on session cookie
324 cookie_samesite =
325 
326 # Default setting for wrap toggle on DAG code and TI log views.
327 default_wrap = False
328 
329 # Send anonymous user activity to your analytics tool
330 # analytics_tool = # choose from google_analytics, segment, or metarouter
331 # analytics_id = XXXXXXXXXXX
332 
333 [email]
334 email_backend = airflow.utils.email.send_email_smtp
335 # 邮件html模板绝对路径位置
336 html_content_template = /mnt/e/airflow_project/airflow_config/local/email_template
337 
338 [smtp]
339 # If you want airflow to send emails on retries, failure, and you want to use
340 # the airflow.utils.email.send_email_smtp function, you have to configure an
341 # smtp server here
342 # 邮件服务 相关配置，根据实际情况配置
343 smtp_host = smtp.exmail.qq.com
344 smtp_starttls = False
345 smtp_ssl = True
346 # Uncomment and set the user/pass settings if you want to use SMTP AUTH
347 smtp_user = [email protected]
348 smtp_password = xxx
349 smtp_port = 465
350 smtp_mail_from = [email protected]
351 
352 
353 [celery]
354 # This section only applies if you are using the CeleryExecutor in
355 # [core] section above
356 
357 # The app name that will be used by celery
358 celery_app_name = airflow.executors.celery_executor
359 
360 # The concurrency that will be used when starting workers with the
361 # "airflow worker" command. This defines the number of task instances that
362 # a worker will take, so size up your workers based on the resources on
363 # your worker box and the nature of your tasks
364 worker_concurrency = 16
365 
366 # The maximum and minimum concurrency that will be used when starting workers with the
367 # "airflow worker" command (always keep minimum processes, but grow to maximum if necessary).
368 # Note the value should be "max_concurrency,min_concurrency"
369 # Pick these numbers based on resources on worker box and the nature of the task.
370 # If autoscale option is available, worker_concurrency will be ignored.
371 # http://docs.celeryproject.org/en/latest/reference/celery.bin.worker.html#cmdoption-celery-worker-autoscale
372 # worker_autoscale = 16,12
373 
374 # When you start an airflow worker, airflow starts a tiny web server
375 # subprocess to serve the workers local log files to the airflow main
376 # web server, who then builds pages and sends them to users. This defines
377 # the port on which the logs are served. It needs to be unused, and open
378 # visible from the main web server to connect into the workers.
379 worker_log_server_port = 8793
380 
381 # The Celery broker URL. Celery supports RabbitMQ, Redis and experimentally
382 # a sqlalchemy database. Refer to the Celery documentation for more
383 # information.
384 # http://docs.celeryproject.org/en/latest/userguide/configuration.html#broker-settings
385 # celery服务 broker连接，此处使用 rabbitmq
386 broker_url = pyamqp://role:[email protected]:5672/
387 
388 # The Celery result_backend. When a job finishes, it needs to update the
389 # metadata of the job. Therefore it will post a message on a message bus,
390 # or insert it into a database (depending of the backend)
391 # This status is used by the scheduler to update the state of the task
392 # The use of a database is highly recommended
393 # http://docs.celeryproject.org/en/latest/userguide/configuration.html#task-result-backend-settings
394 # celery服务 结果存储连接
395 result_backend = redis://localhost/15
396 
397 # Celery Flower is a sweet UI for Celery. Airflow has a shortcut to start
398 # it `airflow flower`. This defines the IP that Celery Flower runs on
399 flower_host = 0.0.0.0
400 
401 # The root URL for Flower
402 # Ex: flower_url_prefix = /flower
403 flower_url_prefix =
404 
405 # This defines the port that Celery Flower runs on
406 flower_port = 5555
407 
408 # Securing Flower with Basic Authentication
409 # Accepts user:password pairs separated by a comma
410 # Example: flower_basic_auth = user1:password1,user2:password2
411 flower_basic_auth =
412 
413 # Default queue that tasks get assigned to and that worker listen on.
414 default_queue = default
415 
416 # How many processes CeleryExecutor uses to sync task state.
417 # 0 means to use max(1, number of cores - 1) processes.
418 sync_parallelism = 0
419 
420 # Import path for celery configuration options
421 celery_config_options = airflow.config_templates.default_celery.DEFAULT_CELERY_CONFIG
422 
423 # In case of using SSL
424 ssl_active = False
425 ssl_key =
426 ssl_cert =
427 ssl_cacert =
428 
429 # Celery Pool implementation.
430 # Choices include: prefork (default), eventlet, gevent or solo.
431 # See:
432 #   https://docs.celeryproject.org/en/latest/userguide/workers.html#concurrency
433 #   https://docs.celeryproject.org/en/latest/userguide/concurrency/eventlet.html
434 pool = prefork
435 
436 [celery_broker_transport_options]
437 # This section is for specifying options which can be passed to the
438 # underlying celery broker transport.  See:
439 # http://docs.celeryproject.org/en/latest/userguide/configuration.html#std:setting-broker_transport_options
440 
441 # The visibility timeout defines the number of seconds to wait for the worker
442 # to acknowledge the task before the message is redelivered to another worker.
443 # Make sure to increase the visibility timeout to match the time of the longest
444 # ETA you're planning to use.
445 #
446 # visibility_timeout is only supported for Redis and SQS celery brokers.
447 # See:
448 #   http://docs.celeryproject.org/en/master/userguide/configuration.html#std:setting-broker_transport_options
449 #
450 #visibility_timeout = 21600
451 
452 [dask]
453 # This section only applies if you are using the DaskExecutor in
454 # [core] section above
455 
456 # The IP address and port of the Dask cluster's scheduler.
457 cluster_address = 127.0.0.1:8786
458 # TLS/ SSL settings to access a secured Dask scheduler.
459 tls_ca =
460 tls_cert =
461 tls_key =
462 
463 
464 [scheduler]
465 # Task instances listen for external kill signal (when you clear tasks
466 # from the CLI or the UI), this defines the frequency at which they should
467 # listen (in seconds).
468 job_heartbeat_sec = 5
469 
470 # The scheduler constantly tries to trigger new tasks (look at the
471 # scheduler section in the docs for more information). This defines
472 # how often the scheduler should run (in seconds).
473 scheduler_heartbeat_sec = 5
474 
475 # after how much time should the scheduler terminate in seconds
476 # -1 indicates to run continuously (see also num_runs)
477 run_duration = -1
478 
479 # after how much time (seconds) a new DAGs should be picked up from the filesystem
480 min_file_process_interval = 0
481 
482 # How often (in seconds) to scan the DAGs directory for new files. Default to 5 minutes.
483 dag_dir_list_interval = 300
484 
485 # How often should stats be printed to the logs
486 print_stats_interval = 30
487 
488 # If the last scheduler heartbeat happened more than scheduler_health_check_threshold ago (in seconds),
489 # scheduler is considered unhealthy.
490 # This is used by the health check in the "/health" endpoint
491 scheduler_health_check_threshold = 30
492 
493 # 定时任务 日志位置
494 child_process_log_directory = /mnt/e/airflow_project/log/airflow/scheduler
495 
496 # Local task jobs periodically heartbeat to the DB. If the job has
497 # not heartbeat in this many seconds, the scheduler will mark the
498 # associated task instance as failed and will re-schedule the task.
499 scheduler_zombie_task_threshold = 300
500 
501 # Turn off scheduler catchup by setting this to False.
502 # Default behavior is unchanged and
503 # Command Line Backfills still work, but the scheduler
504 # will not do scheduler catchup if this is False,
505 # however it can be set on a per DAG basis in the
506 # DAG definition (catchup)
507 catchup_by_default = True
508 
509 # This changes the batch size of queries in the scheduling main loop.
510 # If this is too high, SQL query performance may be impacted by one
511 # or more of the following:
512 #  - reversion to full table scan
513 #  - complexity of query predicate
514 #  - excessive locking
515 #
516 # Additionally, you may hit the maximum allowable query length for your db.
517 #
518 # Set this to 0 for no limit (not advised)
519 max_tis_per_query = 512
520 
521 # Statsd (https://github.com/etsy/statsd) integration settings
522 statsd_on = True
523 statsd_host = localhost
524 statsd_port = 8125
525 statsd_prefix = airflow
526 
527 # The scheduler can run multiple threads in parallel to schedule dags.
528 # This defines how many threads will run.
529 max_threads = 2
530 
531 authenticate = False
532 
533 # Turn off scheduler use of cron intervals by setting this to False.
534 # DAGs submitted manually in the web UI or with trigger_dag will still run.
535 use_job_schedule = True
536 
537 [ldap]
538 # set this to ldaps://:
539 uri =
540 user_filter = objectClass=*
541 user_name_attr = uid
542 group_member_attr = memberOf
543 superuser_filter =
544 data_profiler_filter =
545 bind_user = cn=Manager,dc=example,dc=com
546 bind_password = insecure
547 basedn = dc=example,dc=com
548 cacert = /etc/ca/ldap_ca.crt
549 search_scope = LEVEL
550 
551 # This setting allows the use of LDAP servers that either return a
552 # broken schema, or do not return a schema.
553 ignore_malformed_schema = False
554 
555 [mesos]
556 # Mesos master address which MesosExecutor will connect to.
557 master = localhost:5050
558 
559 # The framework name which Airflow scheduler will register itself as on mesos
560 framework_name = Airflow
561 
562 # Number of cpu cores required for running one task instance using
563 # 'airflow run    --local -p '
564 # command on a mesos slave
565 task_cpu = 1
566 
567 # Memory in MB required for running one task instance using
568 # 'airflow run    --local -p '
569 # command on a mesos slave
570 task_memory = 256
571 
572 # Enable framework checkpointing for mesos
573 # See http://mesos.apache.org/documentation/latest/slave-recovery/
574 checkpoint = False
575 
576 # Failover timeout in milliseconds.
577 # When checkpointing is enabled and this option is set, Mesos waits
578 # until the configured timeout for
579 # the MesosExecutor framework to re-register after a failover. Mesos
580 # shuts down running tasks if the
581 # MesosExecutor framework fails to re-register within this timeframe.
582 # failover_timeout = 604800
583 
584 # Enable framework authentication for mesos
585 # See http://mesos.apache.org/documentation/latest/configuration/
586 authenticate = False
587 
588 # Mesos credentials, if authentication is enabled
589 # default_principal = admin
590 # default_secret = admin
591 
592 # Optional Docker Image to run on slave before running the command
593 # This image should be accessible from mesos slave i.e mesos slave
594 # should be able to pull this docker image before executing the command.
595 # docker_image_slave = puckel/docker-airflow
596 
597 [kerberos]
598 ccache = /tmp/airflow_krb5_ccache
599 # gets augmented with fqdn
600 principal = airflow
601 reinit_frequency = 3600
602 kinit_path = kinit
603 keytab = airflow.keytab
604 
605 
606 [github_enterprise]
607 api_rev = v3
608 
609 [admin]
610 # UI to hide sensitive variable fields when set to True
611 hide_sensitive_variable_fields = True
612 
613 [elasticsearch]
614 # Elasticsearch host
615 host =
616 # Format of the log_id, which is used to query for a given tasks logs
617 log_id_template = {dag_id}-{task_id}-{execution_date}-{try_number}
618 # Used to mark the end of a log stream for a task
619 end_of_log_mark = end_of_log
620 # Qualified URL for an elasticsearch frontend (like Kibana) with a template argument for log_id
621 # Code will construct log_id using the log_id template from the argument above.
622 # NOTE: The code will prefix the https:// automatically, don't include that here.
623 frontend =
624 # Write the task logs to the stdout of the worker, rather than the default files
625 write_stdout = False
626 # Instead of the default log formatter, write the log lines as JSON
627 json_format = False
628 # Log fields to also attach to the json output, if enabled
629 json_fields = asctime, filename, lineno, levelname, message
630 
631 [elasticsearch_configs]
632 
633 use_ssl = False
634 verify_certs = True
635 
636 [kubernetes]
637 # The repository, tag and imagePullPolicy of the Kubernetes Image for the Worker to Run
638 worker_container_repository =
639 worker_container_tag =
640 worker_container_image_pull_policy = IfNotPresent
641 
642 # If True (default), worker pods will be deleted upon termination
643 delete_worker_pods = True
644 
645 # Number of Kubernetes Worker Pod creation calls per scheduler loop
646 worker_pods_creation_batch_size = 1
647 
648 # The Kubernetes namespace where airflow workers should be created. Defaults to `default`
649 namespace = default
650 
651 # The name of the Kubernetes ConfigMap Containing the Airflow Configuration (this file)
652 airflow_configmap =
653 
654 # For docker image already contains DAGs, this is set to `True`, and the worker will search for dags in dags_folder,
655 # otherwise use git sync or dags volume claim to mount DAGs
656 dags_in_image = False
657 
658 # For either git sync or volume mounted DAGs, the worker will look in this subpath for DAGs
659 dags_volume_subpath =
660 
661 # For DAGs mounted via a volume claim (mutually exclusive with git-sync and host path)
662 dags_volume_claim =
663 
664 # For volume mounted logs, the worker will look in this subpath for logs
665 logs_volume_subpath =
666 
667 # A shared volume claim for the logs
668 logs_volume_claim =
669 
670 # For DAGs mounted via a hostPath volume (mutually exclusive with volume claim and git-sync)
671 # Useful in local environment, discouraged in production
672 dags_volume_host =
673 
674 # A hostPath volume for the logs
675 # Useful in local environment, discouraged in production
676 logs_volume_host =
677 
678 # A list of configMapsRefs to envFrom. If more than one configMap is
679 # specified, provide a comma separated list: configmap_a,configmap_b
680 env_from_configmap_ref =
681 
682 # A list of secretRefs to envFrom. If more than one secret is
683 # specified, provide a comma separated list: secret_a,secret_b
684 env_from_secret_ref =
685 
686 # Git credentials and repository for DAGs mounted via Git (mutually exclusive with volume claim)
687 git_repo =
688 git_branch =
689 git_subpath =
690 # Use git_user and git_password for user authentication or git_ssh_key_secret_name and git_ssh_key_secret_key
691 # for SSH authentication
692 git_user =
693 git_password =
694 git_sync_root = /git
695 git_sync_dest = repo
696 # Mount point of the volume if git-sync is being used.
697 # i.e. /Users/wudong/work/Python/flow/dags
698 git_dags_folder_mount_point =
699 
700 # To get Git-sync SSH authentication set up follow this format
701 #
702 # airflow-secrets.yaml:
703 # ---
704 # apiVersion: v1
705 # kind: Secret
706 # metadata:
707 #   name: airflow-secrets
708 # data:
709 #   # key needs to be gitSshKey
710 #   gitSshKey: 
711 # ---
712 # airflow-configmap.yaml:
713 # apiVersion: v1
714 # kind: ConfigMap
715 # metadata:
716 #   name: airflow-configmap
717 # data:
718 #   known_hosts: |
719 #       github.com ssh-rsa <...>
720 #   airflow.cfg: |
721 #       ...
722 #
723 # git_ssh_key_secret_name = airflow-secrets
724 # git_ssh_known_hosts_configmap_name = airflow-configmap
725 git_ssh_key_secret_name =
726 git_ssh_known_hosts_configmap_name =
727 
728 # To give the git_sync init container credentials via a secret, create a secret
729 # with two fields: GIT_SYNC_USERNAME and GIT_SYNC_PASSWORD (example below) and
730 # add `git_sync_credentials_secret = ` to your airflow config under the kubernetes section
731 #
732 # Secret Example:
733 # apiVersion: v1
734 # kind: Secret
735 # metadata:
736 #   name: git-credentials
737 # data:
738 #   GIT_SYNC_USERNAME: 
739 #   GIT_SYNC_PASSWORD: 
740 git_sync_credentials_secret =
741 
742 # For cloning DAGs from git repositories into volumes: https://github.com/kubernetes/git-sync
743 git_sync_container_repository = k8s.gcr.io/git-sync
744 git_sync_container_tag = v3.1.1
745 git_sync_init_container_name = git-sync-clone
746 git_sync_run_as_user = 65533
747 
748 # The name of the Kubernetes service account to be associated with airflow workers, if any.
749 # Service accounts are required for workers that require access to secrets or cluster resources.
750 # See the Kubernetes RBAC documentation for more:
751 #   https://kubernetes.io/docs/admin/authorization/rbac/
752 worker_service_account_name =
753 
754 # Any image pull secrets to be given to worker pods, If more than one secret is
755 # required, provide a comma separated list: secret_a,secret_b
756 image_pull_secrets =
757 
758 # GCP Service Account Keys to be provided to tasks run on Kubernetes Executors
759 # Should be supplied in the format: key-name-1:key-path-1,key-name-2:key-path-2
760 gcp_service_account_keys =
761 
762 # Use the service account kubernetes gives to pods to connect to kubernetes cluster.
763 # It's intended for clients that expect to be running inside a pod running on kubernetes.
764 # It will raise an exception if called from a process not running in a kubernetes environment.
765 in_cluster = True
766 
767 # When running with in_cluster=False change the default cluster_context or config_file
768 # options to Kubernetes client. Leave blank these to use default behaviour like `kubectl` has.
769 # cluster_context =
770 # config_file =
771 
772 
773 # Affinity configuration as a single line formatted JSON object.
774 # See the affinity model for top-level key names (e.g. `nodeAffinity`, etc.):
775 #   https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.12/#affinity-v1-core
776 affinity =
777 
778 # A list of toleration objects as a single line formatted JSON array
779 # See:
780 #   https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.12/#toleration-v1-core
781 tolerations =
782 
783 # **kwargs parameters to pass while calling a kubernetes client core_v1_api methods from Kubernetes Executor
784 # provided as a single line formatted JSON dictionary string.
785 # List of supported params in **kwargs are similar for all core_v1_apis, hence a single config variable for all apis
786 # See:
787 #   https://raw.githubusercontent.com/kubernetes-client/python/master/kubernetes/client/apis/core_v1_api.py
788 # Note that if no _request_timeout is specified, the kubernetes client will wait indefinitely for kubernetes
789 # api responses, which will cause the scheduler to hang. The timeout is specified as [connect timeout, read timeout]
790 kube_client_request_args = {"_request_timeout" : [60,60] }
791 
792 # Worker pods security context options
793 # See:
794 #   https://kubernetes.io/docs/tasks/configure-pod-container/security-context/
795 
796 # Specifies the uid to run the first process of the worker pods containers as
797 run_as_user =
798 
799 # Specifies a gid to associate with all containers in the worker pods
800 # if using a git_ssh_key_secret_name use an fs_group
801 # that allows for the key to be read, e.g. 65533
802 fs_group =
803 
804 [kubernetes_node_selectors]
805 # The Key-value pairs to be given to worker pods.
806 # The worker pods will be scheduled to the nodes of the specified key-value pairs.
807 # Should be supplied in the format: key = value
808 
809 [kubernetes_annotations]
810 # The Key-value annotations pairs to be given to worker pods.
811 # Should be supplied in the format: key = value
812 
813 [kubernetes_environment_variables]
814 # The scheduler sets the following environment variables into your workers. You may define as
815 # many environment variables as needed and the kubernetes launcher will set them in the launched workers.
816 # Environment variables in this section are defined as follows
817 #      = 
818 #
819 # For example if you wanted to set an environment variable with value `prod` and key
820 # `ENVIRONMENT` you would follow the following format:
821 #     ENVIRONMENT = prod
822 #
823 # Additionally you may override worker airflow settings with the AIRFLOW__
__
824 # formatting as supported by airflow normally.
825 
826 [kubernetes_secrets]
827 # The scheduler mounts the following secrets into your workers as they are launched by the
828 # scheduler. You may define as many secrets as needed and the kubernetes launcher will parse the
829 # defined secrets and mount them as secret environment variables in the launched workers.
830 # Secrets in this section are defined as follows
831 #      = =
832 #
833 # For example if you wanted to mount a kubernetes secret key named `postgres_password` from the
834 # kubernetes secret object `airflow-secret` as the environment variable `POSTGRES_PASSWORD` into
835 # your workers you would follow the following format:
836 #     POSTGRES_PASSWORD = airflow-secret=postgres_credentials
837 #
838 # Additionally you may override worker airflow settings with the AIRFLOW__
__
839 # formatting as supported by airflow normally.
840 
841 [kubernetes_labels]
842 # The Key-value pairs to be given to worker pods.
843 # The worker pods will be given these static labels, as well as some additional dynamic labels
844 # to identify the task.
845 # Should be supplied in the format: key = value

View Code

 

错误记录：

* 设置supervisor启动airflow服务时，报错如下

Error: No module named airflow.www.gunicorn_config

* 处理方式

在supervisor的配置文件的 environment常量中添加 PATH="/home/work/www/jerry/venv/bin:%(ENV_PATH)s"

* web界面报错

KeyError: 'Variable xxx does not exist'

* 处理方式

在airflow网页的Admin=>Variables页面添加对应的 变量

 

相关网址：http://airflow.apache.org/index.html