我正在使用集群化的Airflow环境，其中有四个用于服务器的AWS ec2实例。
ec2-instances
到目前为止，我的设置已经可以正常工作三个月，但偶尔会有大约一周一次的时间，当Airflow尝试记录某些内容时，我会收到“管道破裂”的消息。
*** Log file isn't local.
*** Fetching here: http://ip-1-2-3-4:8793/log/foobar/task_1/2018-07-13T00:00:00/1.log
[2018-07-16 00:00:15,521] {cli.py:374} INFO - Running on host ip-1-2-3-4
[2018-07-16 00:00:15,698] {models.py:1197} INFO - Dependencies all met for <TaskInstance: foobar.task_1 2018-07-13 00:00:00 [queued]>
[2018-07-16 00:00:15,710] {models.py:1197} INFO - Dependencies all met for <TaskInstance: foobar.task_1 2018-07-13 00:00:00 [queued]>
[2018-07-16 00:00:15,710] {models.py:1407} INFO -
--------------------------------------------------------------------------------
Starting attempt 1 of 1
--------------------------------------------------------------------------------
[2018-07-16 00:00:15,719] {models.py:1428} INFO - Executing <Task(OmegaFileSensor): task_1> on 2018-07-13 00:00:00
[2018-07-16 00:00:15,720] {base_task_runner.py:115} INFO - Running: ['bash', '-c', 'airflow run foobar task_1 2018-07-13T00:00:00 --job_id 1320 --raw -sd DAGS_FOLDER/datalake_digitalplatform_arl_workflow_schedule_test_2.py']
[2018-07-16 00:00:16,532] {base_task_runner.py:98} INFO - Subtask: [2018-07-16 00:00:16,532] {configuration.py:206} WARNING - section/key [celery/celery_ssl_active] not found in config
[2018-07-16 00:00:16,532] {base_task_runner.py:98} INFO - Subtask: [2018-07-16 00:00:16,532] {default_celery.py:41} WARNING - Celery Executor will run without SSL
[2018-07-16 00:00:16,534] {base_task_runner.py:98} INFO - Subtask: [2018-07-16 00:00:16,533] {__init__.py:45} INFO - Using executor CeleryExecutor
[2018-07-16 00:00:16,597] {base_task_runner.py:98} INFO - Subtask: [2018-07-16 00:00:16,597] {models.py:189} INFO - Filling up the DagBag from /home/ec2-user/airflow/dags/datalake_digitalplatform_arl_workflow_schedule_test_2.py
[2018-07-16 00:00:16,768] {cli.py:374} INFO - Running on host ip-1-2-3-4
[2018-07-16 00:16:24,931] {logging_mixin.py:84} WARNING - --- Logging error ---
[2018-07-16 00:16:24,931] {logging_mixin.py:84} WARNING - Traceback (most recent call last):
[2018-07-16 00:16:24,931] {logging_mixin.py:84} WARNING - File "/usr/lib64/python3.6/logging/__init__.py", line 996, in emit
self.flush()
[2018-07-16 00:16:24,932] {logging_mixin.py:84} WARNING - File "/usr/lib64/python3.6/logging/__init__.py", line 976, in flush
self.stream.flush()
[2018-07-16 00:16:24,932] {logging_mixin.py:84} WARNING - BrokenPipeError: [Errno 32] Broken pipe
[2018-07-16 00:16:24,932] {logging_mixin.py:84} WARNING - Call stack:
[2018-07-16 00:16:24,933] {logging_mixin.py:84} WARNING - File "/usr/bin/airflow", line 27, in <module>
args.func(args)
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - File "/usr/local/lib/python3.6/site-packages/airflow/bin/cli.py", line 392, in run
pool=args.pool,
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - File "/usr/local/lib/python3.6/site-packages/airflow/utils/db.py", line 50, in wrapper
result = func(*args, **kwargs)
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - File "/usr/local/lib/python3.6/site-packages/airflow/models.py", line 1488, in _run_raw_task
result = task_copy.execute(context=context)
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - File "/usr/local/lib/python3.6/site-packages/airflow/operators/sensors.py", line 78, in execute
while not self.poke(context):
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - File "/home/ec2-user/airflow/plugins/custom_plugins.py", line 35, in poke
directory = os.listdir(full_path)
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - File "/usr/local/lib/python3.6/site-packages/airflow/utils/timeout.py", line 36, in handle_timeout
self.log.error("Process timed out")
[2018-07-16 00:16:24,934] {logging_mixin.py:84} WARNING - Message: 'Process timed out'
Arguments: ()
[2018-07-16 00:16:24,942] {models.py:1595} ERROR - Timeout
Traceback (most recent call last):
File "/usr/local/lib/python3.6/site-packages/airflow/models.py", line 1488, in _run_raw_task
result = task_copy.execute(context=context)
File "/usr/local/lib/python3.6/site-packages/airflow/operators/sensors.py", line 78, in execute
while not self.poke(context):
File "/home/ec2-user/airflow/plugins/custom_plugins.py", line 35, in poke
directory = os.listdir(full_path)
File "/usr/local/lib/python3.6/site-packages/airflow/utils/timeout.py", line 37, in handle_timeout
raise AirflowTaskTimeout(self.error_message)
airflow.exceptions.AirflowTaskTimeout: Timeout
[2018-07-16 00:16:24,942] {models.py:1624} INFO - Marking task as FAILED.
[2018-07-16 00:16:24,956] {models.py:1644} ERROR - Timeout
有时错误还会提示
*** Log file isn't local.
*** Fetching here: http://ip-1-2-3-4:8793/log/foobar/task_1/2018-07-12T00:00:00/1.log
*** Failed to fetch log file from worker. 404 Client Error: NOT FOUND for url: http://ip-1-2-3-4:8793/log/foobar/task_1/2018-07-12T00:00:00/1.log
我不确定为什么日志在大约95％的时间内都能正常工作，而在其他时间却随机出现故障。这是我Airflow.cfg文件中的日志设置，
# The folder where airflow should store its log files
# This path must be absolute
base_log_folder = /home/ec2-user/airflow/logs
# Airflow can store logs remotely in AWS S3 or Google Cloud Storage. Users
# must supply an Airflow connection id that provides access to the storage
# location.
remote_log_conn_id =
encrypt_s3_logs = False
# Logging level
logging_level = INFO
# Logging class
# Specify the class that will specify the logging configuration
# This class has to be on the python classpath
# logging_config_class = my.path.default_local_settings.LOGGING_CONFIG
logging_config_class =
# Log format
log_format = [%%(asctime)s] {%%(filename)s:%%(lineno)d} %%(levelname)s - %%(message)s
simple_log_format = %%(asctime)s %%(levelname)s - %%(message)s
# Name of handler to read task instance logs.
# Default to use file task handler.
task_log_reader = file.task
# Log files for the gunicorn webserver. '-' means log to stderr.
access_logfile = -
error_logfile =
# The amount of time (in secs) webserver will wait for initial handshake
# while fetching logs from other worker machine
log_fetch_timeout_sec = 5
# When you start an airflow worker, airflow starts a tiny web server
# subprocess to serve the workers local log files to the airflow main
# web server, who then builds pages and sends them to users. This defines
# the port on which the logs are served. It needs to be unused, and open
# visible from the main web server to connect into the workers.
worker_log_server_port = 8793
# How often should stats be printed to the logs
print_stats_interval = 30
child_process_log_directory = /home/ec2-user/airflow/logs/scheduler
我想知道是否应该尝试其他记录技术，例如写入S3存储桶，或者是否有其他方法可以解决此问题。
更新：
将日志写入S3不能解决此问题。而且，该错误现在更加一致（仍然是零星的）。现在大约有50％的时间在发生。我注意到的一件事是发生的任务是我的AWS EMR创建任务。启动AWS EMR集群大约需要20分钟，然后任务必须等待Spark命令在EMR集群上运行。因此，单个任务运行大约30分钟。我想知道这是否太长时间以至于无法运行Airflow任务，这就是为什么它开始无法写入日志的原因。如果是这种情况，那么我可以分解EMR任务，以便在EMR集群上有一个创建EMR的任务，然后是另一个针对Spark命令的任务。
注意：
上在Airflow的Jira上创建了新的故障单答案 0 :(得分：2)
这个问题是我刚刚在这里AirflowException: Celery command failed - The recorded hostname does not match this instance's hostname解决的另一个问题的症状。
我有一段时间没有看到 AirflowException：Celery命令失败，因为它出现在 airflow worker 日志中。直到我看到气流工作人员实时记录日志后，我才发现抛出错误时，我的任务中也出现了BrokenPipeException。
虽然有点奇怪。如果在工作节点上发生print("something to log") 和 AirflowException: Celery command failed...错误，我只会看到BrokenPipeException抛出。当我将所有打印语句更改为使用import logging ... logging.info("something to log")时，我不会看到BrokenPipeException ，但是该任务仍然会由于AirflowException: Celery command failed...错误而失败。但是如果我没有在Airflow任务日志中看到BrokenPipeException被抛出，我将不知道为什么任务失败了，因为一旦我消除了打印语句，我就再也没有在Airflow任务日志中看到任何错误（仅在 $气流工人日志）
长话短说，有很多收获。
不要通过导入日志记录然后使用print("something to log")然后使用import logging
如果您将AWS EC2-Instance用作Airflow的服务器，则可能遇到此问题：https://github.com/apache/incubator-airflow/pull/2484该问题的修复已集成到Airflow版本1.10中（I' m当前使用Airflow版本1.9）。因此，升级您的Airflow version to 1.10。您也可以使用the command here logging.info("something to log")。另外，如果您不想升级Airflow版本，则可以按照the github issue上的步骤操作，以使用修复程序手动更新文件，或者派生Airflow并选择修复该文件的提交。