I need to run a websocket server on ESP32 and the official example raises the following exception when I connect from any client:
MPY: soft reboot
Network config: ('192.168.0.200', '255.255.255.0', '192.168.0.1', '8.8.8.8')
b'Sec-WebSocket-Version: 13\r\n'
b'Sec-WebSocket-Key: k5Lr79cZgBQg7irI247FMw==\r\n'
b'Connection: Upgrade\r\n'
b'Upgrade: websocket\r\n'
b'Sec-WebSocket-Extensions: permessage-deflate; client_max_window_bits\r\n'
b'Host: 192.168.0.200\r\n'
b'\r\n'
Finished webrepl handshake
Task exception wasn't retrieved
future: <Task> coro= <generator object 'echo' at 3ffe79b0>
Traceback (most recent call last):
File "uasyncio/core.py", line 1, in run_until_complete
File "main.py", line 22, in echo
File "uasyncio/websocket/server.py", line 60, in WSReader
AttributeError: 'Stream' object has no attribute 'ios'
My micropython firmware and libraries:
Micropython firmware: https://micropython.org/resources/firmware/esp32-idf3-20200902-v1.13.bin
Pip libraries installed: micropython-ulogging, uasyncio.websocket.server
My main.py:
import network
import machine
sta_if = network.WLAN(network.STA_IF)
sta_if.active(True)
sta_if.ifconfig(('192.168.0.200', '255.255.255.0', '192.168.0.1', '8.8.8.8'))
if not sta_if.isconnected():
print('connecting to network...')
sta_if.connect('my-ssid', 'my-password')
while not sta_if.isconnected():
machine.idle() # save power while waiting
print('Network config:', sta_if.ifconfig())
# from https://github.com/micropython/micropython-lib/blob/master/uasyncio.websocket.server/example_websock.py
import uasyncio
from uasyncio.websocket.server import WSReader, WSWriter
def echo(reader, writer):
# Consume GET line
yield from reader.readline()
reader = yield from WSReader(reader, writer)
writer = WSWriter(reader, writer)
while 1:
l = yield from reader.read(256)
print(l)
if l == b"\r":
await writer.awrite(b"\r\n")
else:
await writer.awrite(l)
import ulogging as logging
#logging.basicConfig(level=logging.INFO)
logging.basicConfig(level=logging.DEBUG)
loop = uasyncio.get_event_loop()
loop.create_task(uasyncio.start_server(echo, "0.0.0.0", 80))
loop.run_forever()
loop.close()
MicroPython 1.13 implements asyncio v3 which has a number of breaking changes compared to the 3 year old sample referenced.
I suggest you refer to Peter Hinch's excellent documentation on asyncio,
and the asyncio V3 tutorial
I encountered the same problem. I looked at the old implementation of Stream class [1] and the new one [2].
It seems to me, that you need to edit server.py from uasyncio/websocket/.
You can download the files from [3] to your PC. Then at the bottom of the file replace the two instances of "reader.ios" by "reader.s".
Save the file to your ESP32 and it should work. Of course you need to use "from server import WSReader, WSWriter" instead of "from uasyncio.websocket.server import WSReader, WSWriter".
[1] https://github.com/pfalcon/pycopy-lib/blob/master/uasyncio/uasyncio/__init__.py
[2] https://github.com/micropython/micropython/blob/master/extmod/uasyncio/stream.py
[3] https://pypi.org/project/micropython-uasyncio.websocket.server/#files
https://github.com/pfalcon/pycopy-lib/tree/master/uasyncio has a recent (may'21) sample that should also work on standard MicroPython.
or checkout https://awesome-micropython.com under web servers
I am working on a design pattern to make my python unittest as a POM, so far I have written my page classes in modules HomePageObject.py,FilterPageObject.py, my base class (for common stuff)TestBase in BaseTest.py, my testcase modules are TestCase1.py and TestCase2.py and one runner module runner.py.
In runner class i am using loader.getTestCaseNames to get all the tests from a testcase class of a module. In both the testcase modules the name of the test class is same 'Test' and also the method name is same 'testName'
Since the names are confilicting while importing it in runner, only one test is getting executed. I want python to scan all the modules that i specify for tests in them and run those even the name of classes are same.
I got to know that nose might be helpful in this, but not sure how i can implement it here. Any advice ?
BaseTest.py
from selenium import webdriver
from selenium.webdriver.support.wait import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver import ChromeOptions
import unittest
class TestBase(unittest.TestCase):
driver = None
def __init__(self,testName,browser):
self.browser = browser
super(TestBase,self).__init__(testName)
def setUp(self):
if self.browser == "firefox":
TestBase.driver = webdriver.Firefox()
elif self.browser == "chrome":
options = ChromeOptions()
options.add_argument("--start-maximized")
TestBase.driver = webdriver.Chrome(chrome_options=options)
self.url = "https://www.airbnb.co.in/"
self.driver = TestBase.getdriver()
TestBase.driver.implicitly_wait(10)
def tearDown(self):
self.driver.quit()
#staticmethod
def getdriver():
return TestBase.driver
#staticmethod
def waitForElementVisibility(locator, expression, message):
try:
WebDriverWait(TestBase.driver, 20).\
until(EC.presence_of_element_located((locator, expression)),
message)
return True
except:
return False
TestCase1.py and TestCase2.py (same)
from airbnb.HomePageObject import HomePage
from airbnb.BaseTest import TestBase
class Test(TestBase):
def __init__(self,testName,browser):
super(Test,self).__init__(testName,browser)
def testName(self):
try:
self.driver.get(self.url)
h_page = HomePage()
f_page = h_page.seachPlace("Sicily,Italy")
f_page.selectExperience()
finally:
self.driver.quit()
runner.py
import unittest
from airbnb.TestCase1 import Test
from airbnb.TestCase2 import Test
loader = unittest.TestLoader()
test_names = loader.getTestCaseNames(Test)
suite = unittest.TestSuite()
for test in test_names:
suite.addTest(Test(test,"chrome"))
runner = unittest.TextTestRunner()
result = runner.run(suite)
Also even that one test case is getting passed, some error message is coming
Ran 1 test in 9.734s
OK
Traceback (most recent call last):
File "F:\eclipse-jee-neon-3-win32\eclipse\plugins\org.python.pydev.core_6.3.3.201805051638\pysrc\runfiles.py", line 275, in <module>
main()
File "F:\eclipse-jee-neon-3-win32\eclipse\plugins\org.python.pydev.core_6.3.3.201805051638\pysrc\runfiles.py", line 97, in main
return pydev_runfiles.main(configuration) # Note: still doesn't return a proper value.
File "F:\eclipse-jee-neon-3-win32\eclipse\plugins\org.python.pydev.core_6.3.3.201805051638\pysrc\_pydev_runfiles\pydev_runfiles.py", line 874, in main
PydevTestRunner(configuration).run_tests()
File "F:\eclipse-jee-neon-3-win32\eclipse\plugins\org.python.pydev.core_6.3.3.201805051638\pysrc\_pydev_runfiles\pydev_runfiles.py", line 773, in run_tests
all_tests = self.find_tests_from_modules(file_and_modules_and_module_name)
File "F:\eclipse-jee-neon-3-win32\eclipse\plugins\org.python.pydev.core_6.3.3.201805051638\pysrc\_pydev_runfiles\pydev_runfiles.py", line 629, in find_tests_from_modules
suite = loader.loadTestsFromModule(m)
File "C:\Python27\lib\unittest\loader.py", line 65, in loadTestsFromModule
tests.append(self.loadTestsFromTestCase(obj))
File "C:\Python27\lib\unittest\loader.py", line 56, in loadTestsFromTestCase
loaded_suite = self.suiteClass(map(testCaseClass, testCaseNames))
TypeError: __init__() takes exactly 3 arguments (2 given)
I did this by searching for all the modules of test classes with a pattern and then used __import__(modulename) and called its Test class with desired parameters,
Here is my runner.py
import unittest
import glob
loader = unittest.TestLoader()
suite = unittest.TestSuite()
test_file_strings = glob.glob('Test*.py')
module_strings = [str[0:len(str)-3] for str in test_file_strings]
for module in module_strings:
mod = __import__(module)
test_names =loader.getTestCaseNames(mod.Test)
for test in test_names:
suite.addTest(mod.Test(test,"chrome"))
runner = unittest.TextTestRunner()
result = runner.run(suite)
This worked but still looking for some organized solutions.
(Not sure why second time its showing Ran 0 tests in 0.000s )
Finding files... done.
Importing test modules ... ..done.
----------------------------------------------------------------------
Ran 2 tests in 37.491s
OK
----------------------------------------------------------------------
Ran 0 tests in 0.000s
OK
I'm trying to send data from my dronekit.io vehicle using flask-socket.io. Unfortunately, I got this log:
Starting copter simulator (SITL)
SITL already Downloaded and Extracted.
Ready to boot.
Connecting to vehicle on: tcp:127.0.0.1:5760
>>> APM:Copter V3.3 (d6053245)
>>> Frame: QUAD
>>> Calibrating barometer
>>> Initialising APM...
>>> barometer calibration complete
>>> GROUND START
* Restarting with stat
latitude -35.363261
>>> Exception in attribute handler for location.global_relative_frame
>>> Working outside of request context.
This typically means that you attempted to use functionality that needed
an active HTTP request. Consult the documentation on testing for
information about how to avoid this problem.
longitude 149.1652299
>>> Exception in attribute handler for location.global_relative_frame
>>> Working outside of request context.
This typically means that you attempted to use functionality that needed
an active HTTP request. Consult the documentation on testing for
information about how to avoid this problem.
Here is my code:
sample.py
from dronekit import connect, VehicleMode
from flask import Flask
from flask_socketio import SocketIO, emit
import dronekit_sitl
import time
sitl = dronekit_sitl.start_default()
connection_string = sitl.connection_string()
print("Connecting to vehicle on: %s" % (connection_string,))
vehicle = connect(connection_string, wait_ready=True)
def arm_and_takeoff(aTargetAltitude):
print "Basic pre-arm checks"
while not vehicle.is_armable:
print " Waiting for vehicle to initialise..."
time.sleep(1)
print "Arming motors"
vehicle.mode = VehicleMode("GUIDED")
vehicle.armed = True
while not vehicle.armed:
print " Waiting for arming..."
time.sleep(1)
print "Taking off!"
vehicle.simple_takeoff(aTargetAltitude)
while True:
if vehicle.location.global_relative_frame.alt>=aTargetAltitude*0.95:
print "Reached target altitude"
break
time.sleep(1)
last_latitude = 0.0
last_longitude = 0.0
last_altitude = 0.0
#vehicle.on_attribute('location.global_relative_frame')
def location_callback(self, attr_name, value):
global last_latitude
global last_longitude
global last_altitude
if round(value.lat, 6) != round(last_latitude, 6):
last_latitude = value.lat
print "latitude ", value.lat, "\n"
emit("latitude", value.lat)
if round(value.lon, 6) != round(last_longitude, 6):
last_longitude = value.lon
print "longitude ", value.lon, "\n"
emit("longitude", value.lon)
if round(value.alt) != round(last_altitude):
last_altitude = value.alt
print "altitude ", value.alt, "\n"
emit("altitude", value.alt)
app = Flask(__name__)
socketio = SocketIO(app)
if __name__ == '__main__':
socketio.run(app, host='0.0.0.0', port=5000, debug=True)
arm_and_takeoff(20)
I know because of the logs that I should not do any HTTP request inside "vehicle.on_attribute" decorator method and I should search for information on how to solve this problem but I didn't found any info about the error.
Hope you could help me.
Thank you very much,
Raniel
The emit() function by default returns an event back to the active client. If you call this function outside of a request context there is no concept of active client, so you get this error.
You have a couple of options:
indicate the recipient of the event and the namespace that you are using, so that there is no need to look them up in the context. You can do this by adding room and namespace arguments. Use '/' for the namespace if you are using the default namespace.
emit to all clients by adding broadcast=True as an argument, plus the namespace as indicated in #1.
I'm running 5 DAG's which have generated a total of about 6GB of log data in the base_log_folder over a months period. I just added a remote_base_log_folder but it seems it does not exclude logging to the base_log_folder.
Is there anyway to automatically remove old log files, rotate them or force airflow to not log on disk (base_log_folder) only in remote storage?
Please refer https://github.com/teamclairvoyant/airflow-maintenance-dags
This plugin has DAGs that can kill halted tasks and log-cleanups.
You can grab the concepts and can come up with a new DAG that can cleanup as per your requirement.
We remove the Task logs by implementing our own FileTaskHandler, and then pointing to it in the airflow.cfg. So, we overwrite the default LogHandler to keep only N task logs, without scheduling additional DAGs.
We are using Airflow==1.10.1.
[core]
logging_config_class = log_config.LOGGING_CONFIG
log_config.LOGGING_CONFIG
BASE_LOG_FOLDER = conf.get('core', 'BASE_LOG_FOLDER')
FOLDER_TASK_TEMPLATE = '{{ ti.dag_id }}/{{ ti.task_id }}'
FILENAME_TEMPLATE = '{{ ti.dag_id }}/{{ ti.task_id }}/{{ ts }}/{{ try_number }}.log'
LOGGING_CONFIG = {
'formatters': {},
'handlers': {
'...': {},
'task': {
'class': 'file_task_handler.FileTaskRotationHandler',
'formatter': 'airflow.job',
'base_log_folder': os.path.expanduser(BASE_LOG_FOLDER),
'filename_template': FILENAME_TEMPLATE,
'folder_task_template': FOLDER_TASK_TEMPLATE,
'retention': 20
},
'...': {}
},
'loggers': {
'airflow.task': {
'handlers': ['task'],
'level': JOB_LOG_LEVEL,
'propagate': False,
},
'airflow.task_runner': {
'handlers': ['task'],
'level': LOG_LEVEL,
'propagate': True,
},
'...': {}
}
}
file_task_handler.FileTaskRotationHandler
import os
import shutil
from airflow.utils.helpers import parse_template_string
from airflow.utils.log.file_task_handler import FileTaskHandler
class FileTaskRotationHandler(FileTaskHandler):
def __init__(self, base_log_folder, filename_template, folder_task_template, retention):
"""
:param base_log_folder: Base log folder to place logs.
:param filename_template: template filename string.
:param folder_task_template: template folder task path.
:param retention: Number of folder logs to keep
"""
super(FileTaskRotationHandler, self).__init__(base_log_folder, filename_template)
self.retention = retention
self.folder_task_template, self.folder_task_template_jinja_template = \
parse_template_string(folder_task_template)
#staticmethod
def _get_directories(path='.'):
return next(os.walk(path))[1]
def _render_folder_task_path(self, ti):
if self.folder_task_template_jinja_template:
jinja_context = ti.get_template_context()
return self.folder_task_template_jinja_template.render(**jinja_context)
return self.folder_task_template.format(dag_id=ti.dag_id, task_id=ti.task_id)
def _init_file(self, ti):
relative_path = self._render_folder_task_path(ti)
folder_task_path = os.path.join(self.local_base, relative_path)
subfolders = self._get_directories(folder_task_path)
to_remove = set(subfolders) - set(subfolders[-self.retention:])
for dir_to_remove in to_remove:
full_dir_to_remove = os.path.join(folder_task_path, dir_to_remove)
print('Removing', full_dir_to_remove)
shutil.rmtree(full_dir_to_remove)
return FileTaskHandler._init_file(self, ti)
Airflow maintainers don't think truncating logs is a part of airflow core logic, to see this, and then in this issue, maintainers suggest to change LOG_LEVEL avoid too many log data.
And in this PR, we can learn how to change log level in airflow.cfg.
good luck.
I know it sounds savage, but have you tried pointing base_log_folder to /dev/null? I use Airflow as a part of a container, so I don't care about the files either, as long as the logger pipe to STDOUT as well.
Not sure how well this plays with S3 though.
For your concrete problems, I have some suggestions.
For those, you would always need a specialized logging config as described in this answer: https://stackoverflow.com/a/54195537/2668430
automatically remove old log files and rotate them
I don't have any practical experience with the TimedRotatingFileHandler from the Python standard library yet, but you might give it a try:
https://docs.python.org/3/library/logging.handlers.html#timedrotatingfilehandler
It not only offers to rotate your files based on a time interval, but if you specify the backupCount parameter, it even deletes your old log files:
If backupCount is nonzero, at most backupCount files will be kept, and if more would be created when rollover occurs, the oldest one is deleted. The deletion logic uses the interval to determine which files to delete, so changing the interval may leave old files lying around.
Which sounds pretty much like the best solution for your first problem.
force airflow to not log on disk (base_log_folder), but only in remote storage?
In this case you should specify the logging config in such a way that you do not have any logging handlers that write to a file, i.e. remove all FileHandlers.
Rather, try to find logging handlers that send the output directly to a remote address.
E.g. CMRESHandler which logs directly to ElasticSearch but needs some extra fields in the log calls.
Alternatively, write your own handler class and let it inherit from the Python standard library's HTTPHandler.
A final suggestion would be to combine both the TimedRotatingFileHandler and setup ElasticSearch together with FileBeat, so you would be able to store your logs inside ElasticSearch (i.e. remote), but you wouldn't store a huge amount of logs on your Airflow disk since they will be removed by the backupCount retention policy of your TimedRotatingFileHandler.
Usually apache airflow grab the disk space due to 3 reasons
1. airflow scheduler logs files
2. mysql binaly logs [Major]
3. xcom table records.
To make it clean up on regular basis I have set up a dag which run on daily basis and cleans the binary logs and truncate the xcom table to make the disk space free
You also might need to install [pip install mysql-connector-python].
To clean up scheduler log files I do delete them manually two times in a week to avoid the risk of logs deleted which needs to be required for some reasons.
I clean the logs files by [sudo rm -rd airflow/logs/] command.
Below is my python code for reference
'
"""Example DAG demonstrating the usage of the PythonOperator."""
from airflow import DAG
from airflow.operators.python import PythonOperator
from datetime import datetime, timedelta
from airflow.utils.dates import days_ago
from airflow.operators.bash import BashOperator
from airflow.providers.postgres.operators.postgres import PostgresOperator
args = {
'owner': 'airflow',
'email_on_failure':True,
'retries': 1,
'email':['Your Email Id'],
'retry_delay': timedelta(minutes=5)
}
dag = DAG(
dag_id='airflow_logs_cleanup',
default_args=args,
schedule_interval='#daily',
start_date=days_ago(0),
catchup=False,
max_active_runs=1,
tags=['airflow_maintenance'],
)
def truncate_table():
import mysql.connector
connection = mysql.connector.connect(host='localhost',
database='db_name',
user='username',
password='your password',
auth_plugin='mysql_native_password')
cursor = connection.cursor()
sql_select_query = """TRUNCATE TABLE xcom"""
cursor.execute(sql_select_query)
connection.commit()
connection.close()
print("XCOM Table truncated successfully")
def delete_binary_logs():
import mysql.connector
from datetime import datetime
date = datetime.today().strftime('%Y-%m-%d')
connection = mysql.connector.connect(host='localhost',
database='db_name',
user='username',
password='your_password',
auth_plugin='mysql_native_password')
cursor = connection.cursor()
query = 'PURGE BINARY LOGS BEFORE ' + "'" + str(date) + "'"
sql_select_query = query
cursor.execute(sql_select_query)
connection.commit()
connection.close()
print("Binary logs deleted successfully")
t1 = PythonOperator(
task_id='truncate_table',
python_callable=truncate_table, dag=dag
)
t2 = PythonOperator(
task_id='delete_binary_logs',
python_callable=delete_binary_logs, dag=dag
)
t2 << t1
'
I am surprized but it worked for me. Update your config as below:
base_log_folder=""
It is test in minio and in s3.
Our solution looks a lot like Franzi's:
Running on Airflow 2.0.1 (py3.8)
Override default logging configuration
Since we use a helm chart for airflow deployment it was easiest to push an env there, but it can also be done in the airflow.cfg or using ENV in dockerfile.
# Set custom logging configuration to enable log rotation for task logging
AIRFLOW__LOGGING__LOGGING_CONFIG_CLASS: "airflow_plugins.settings.airflow_local_settings.DEFAULT_LOGGING_CONFIG"
Then we added the logging configuration together with the custom log handler to a python module we build and install in the docker image. As described here: https://airflow.apache.org/docs/apache-airflow/stable/modules_management.html
Logging configuration snippet
This is only a copy on the default from the airflow codebase, but then the task logger gets a different handler.
DEFAULT_LOGGING_CONFIG: Dict[str, Any] = {
'version': 1,
'disable_existing_loggers': False,
'formatters': {
'airflow': {'format': LOG_FORMAT},
'airflow_coloured': {
'format': COLORED_LOG_FORMAT if COLORED_LOG else LOG_FORMAT,
'class': COLORED_FORMATTER_CLASS if COLORED_LOG else 'logging.Formatter',
},
},
'handlers': {
'console': {
'class': 'airflow.utils.log.logging_mixin.RedirectStdHandler',
'formatter': 'airflow_coloured',
'stream': 'sys.stdout',
},
'task': {
'class': 'airflow_plugins.log.rotating_file_task_handler.RotatingFileTaskHandler',
'formatter': 'airflow',
'base_log_folder': os.path.expanduser(BASE_LOG_FOLDER),
'filename_template': FILENAME_TEMPLATE,
'maxBytes': 10485760, # 10MB
'backupCount': 6,
},
...
RotatingFileTaskHandler
And finally the custom handler which is just a merge of the logging.handlers.RotatingFileHandler and the FileTaskHandler.
#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
"""File logging handler for tasks."""
import logging
import os
from pathlib import Path
from typing import TYPE_CHECKING, Optional
import requests
from airflow.configuration import AirflowConfigException, conf
from airflow.utils.helpers import parse_template_string
if TYPE_CHECKING:
from airflow.models import TaskInstance
class RotatingFileTaskHandler(logging.Handler):
"""
FileTaskHandler is a python log handler that handles and reads
task instance logs. It creates and delegates log handling
to `logging.FileHandler` after receiving task instance context.
It reads logs from task instance's host machine.
:param base_log_folder: Base log folder to place logs.
:param filename_template: template filename string
"""
def __init__(self, base_log_folder: str, filename_template: str, maxBytes=0, backupCount=0):
self.max_bytes = maxBytes
self.backup_count = backupCount
super().__init__()
self.handler = None # type: Optional[logging.FileHandler]
self.local_base = base_log_folder
self.filename_template, self.filename_jinja_template = parse_template_string(filename_template)
def set_context(self, ti: "TaskInstance"):
"""
Provide task_instance context to airflow task handler.
:param ti: task instance object
"""
local_loc = self._init_file(ti)
self.handler = logging.handlers.RotatingFileHandler(
filename=local_loc,
mode='a',
maxBytes=self.max_bytes,
backupCount=self.backup_count,
encoding='utf-8',
delay=False,
)
if self.formatter:
self.handler.setFormatter(self.formatter)
self.handler.setLevel(self.level)
def emit(self, record):
if self.handler:
self.handler.emit(record)
def flush(self):
if self.handler:
self.handler.flush()
def close(self):
if self.handler:
self.handler.close()
def _render_filename(self, ti, try_number):
if self.filename_jinja_template:
if hasattr(ti, 'task'):
jinja_context = ti.get_template_context()
jinja_context['try_number'] = try_number
else:
jinja_context = {
'ti': ti,
'ts': ti.execution_date.isoformat(),
'try_number': try_number,
}
return self.filename_jinja_template.render(**jinja_context)
return self.filename_template.format(
dag_id=ti.dag_id,
task_id=ti.task_id,
execution_date=ti.execution_date.isoformat(),
try_number=try_number,
)
def _read_grouped_logs(self):
return False
def _read(self, ti, try_number, metadata=None): # pylint: disable=unused-argument
"""
Template method that contains custom logic of reading
logs given the try_number.
:param ti: task instance record
:param try_number: current try_number to read log from
:param metadata: log metadata,
can be used for steaming log reading and auto-tailing.
:return: log message as a string and metadata.
"""
# Task instance here might be different from task instance when
# initializing the handler. Thus explicitly getting log location
# is needed to get correct log path.
log_relative_path = self._render_filename(ti, try_number)
location = os.path.join(self.local_base, log_relative_path)
log = ""
if os.path.exists(location):
try:
with open(location) as file:
log += f"*** Reading local file: {location}\n"
log += "".join(file.readlines())
except Exception as e: # pylint: disable=broad-except
log = f"*** Failed to load local log file: {location}\n"
log += "*** {}\n".format(str(e))
elif conf.get('core', 'executor') == 'KubernetesExecutor': # pylint: disable=too-many-nested-blocks
try:
from airflow.kubernetes.kube_client import get_kube_client
kube_client = get_kube_client()
if len(ti.hostname) >= 63:
# Kubernetes takes the pod name and truncates it for the hostname. This truncated hostname
# is returned for the fqdn to comply with the 63 character limit imposed by DNS standards
# on any label of a FQDN.
pod_list = kube_client.list_namespaced_pod(conf.get('kubernetes', 'namespace'))
matches = [
pod.metadata.name
for pod in pod_list.items
if pod.metadata.name.startswith(ti.hostname)
]
if len(matches) == 1:
if len(matches[0]) > len(ti.hostname):
ti.hostname = matches[0]
log += '*** Trying to get logs (last 100 lines) from worker pod {} ***\n\n'.format(
ti.hostname
)
res = kube_client.read_namespaced_pod_log(
name=ti.hostname,
namespace=conf.get('kubernetes', 'namespace'),
container='base',
follow=False,
tail_lines=100,
_preload_content=False,
)
for line in res:
log += line.decode()
except Exception as f: # pylint: disable=broad-except
log += '*** Unable to fetch logs from worker pod {} ***\n{}\n\n'.format(ti.hostname, str(f))
else:
url = os.path.join("http://{ti.hostname}:{worker_log_server_port}/log", log_relative_path).format(
ti=ti, worker_log_server_port=conf.get('celery', 'WORKER_LOG_SERVER_PORT')
)
log += f"*** Log file does not exist: {location}\n"
log += f"*** Fetching from: {url}\n"
try:
timeout = None # No timeout
try:
timeout = conf.getint('webserver', 'log_fetch_timeout_sec')
except (AirflowConfigException, ValueError):
pass
response = requests.get(url, timeout=timeout)
response.encoding = "utf-8"
# Check if the resource was properly fetched
response.raise_for_status()
log += '\n' + response.text
except Exception as e: # pylint: disable=broad-except
log += "*** Failed to fetch log file from worker. {}\n".format(str(e))
return log, {'end_of_log': True}
def read(self, task_instance, try_number=None, metadata=None):
"""
Read logs of given task instance from local machine.
:param task_instance: task instance object
:param try_number: task instance try_number to read logs from. If None
it returns all logs separated by try_number
:param metadata: log metadata,
can be used for steaming log reading and auto-tailing.
:return: a list of listed tuples which order log string by host
"""
# Task instance increments its try number when it starts to run.
# So the log for a particular task try will only show up when
# try number gets incremented in DB, i.e logs produced the time
# after cli run and before try_number + 1 in DB will not be displayed.
if try_number is None:
next_try = task_instance.next_try_number
try_numbers = list(range(1, next_try))
elif try_number < 1:
logs = [
[('default_host', f'Error fetching the logs. Try number {try_number} is invalid.')],
]
return logs, [{'end_of_log': True}]
else:
try_numbers = [try_number]
logs = [''] * len(try_numbers)
metadata_array = [{}] * len(try_numbers)
for i, try_number_element in enumerate(try_numbers):
log, metadata = self._read(task_instance, try_number_element, metadata)
# es_task_handler return logs grouped by host. wrap other handler returning log string
# with default/ empty host so that UI can render the response in the same way
logs[i] = log if self._read_grouped_logs() else [(task_instance.hostname, log)]
metadata_array[i] = metadata
return logs, metadata_array
def _init_file(self, ti):
"""
Create log directory and give it correct permissions.
:param ti: task instance object
:return: relative log path of the given task instance
"""
# To handle log writing when tasks are impersonated, the log files need to
# be writable by the user that runs the Airflow command and the user
# that is impersonated. This is mainly to handle corner cases with the
# SubDagOperator. When the SubDagOperator is run, all of the operators
# run under the impersonated user and create appropriate log files
# as the impersonated user. However, if the user manually runs tasks
# of the SubDagOperator through the UI, then the log files are created
# by the user that runs the Airflow command. For example, the Airflow
# run command may be run by the `airflow_sudoable` user, but the Airflow
# tasks may be run by the `airflow` user. If the log files are not
# writable by both users, then it's possible that re-running a task
# via the UI (or vice versa) results in a permission error as the task
# tries to write to a log file created by the other user.
relative_path = self._render_filename(ti, ti.try_number)
full_path = os.path.join(self.local_base, relative_path)
directory = os.path.dirname(full_path)
# Create the log file and give it group writable permissions
# TODO(aoen): Make log dirs and logs globally readable for now since the SubDag
# operator is not compatible with impersonation (e.g. if a Celery executor is used
# for a SubDag operator and the SubDag operator has a different owner than the
# parent DAG)
Path(directory).mkdir(mode=0o777, parents=True, exist_ok=True)
if not os.path.exists(full_path):
open(full_path, "a").close()
# TODO: Investigate using 444 instead of 666.
os.chmod(full_path, 0o666)
return full_path
Maybe a final note; the links in the airflow UI to the logging will now only open the latest logfile, not the older rotated files which are only accessible by means of SSH or any other interface to access the airflow logging path.
I don't think that there is a rotation mechanism but you can store them in S3 or google cloud storage as describe here : https://airflow.incubator.apache.org/configuration.html#logs