SOLUTIONS POSTED BELOW - TLDR; this dumbs down to using presence_of_element_located with the wrong use case and then not thinking of browser and driver desync.
ISSUE: Issue in which tests pass locally but fail via Jenkins.
Local always passes (windows VM)
Jenkins sometimes passes (windows VM)
Both the local tests and Jenkins tests are run on the same VM (Windows agent). I have followed the advice from these people by adding the following arguments with no change/success:
options = Options()
options.add_experimental_option('excludeSwitches', ['enable-logging']) #Fixes the problem with console outputing warning "Bluetooth: bluetooth_adapter_winrt.cc:1074 Getting Default Adapter failed."
options.add_argument('--no-sandbox')
options.add_argument("enable-automation")
#options.add_argument("--headless") #CAUSES TIMEOUT ERROR
options.add_argument("--window-size=1920,1080")
options.add_argument("--disable-extensions")
options.add_argument("--dns-prefetch-disable")
options.add_argument("--disable-gpu")
options.add_argument("--disable-dev-shm-usage")
options.add_argument("--disable-browser-side-navigation")
options.add_argument("--force-device-scale-factor=1")
prefs = {"download.default_directory" : r"C:\Users\MyUser\Path\Directory\\"}
options.add_experimental_option("prefs", prefs)
s = Service('chromeDriverPath/chromedriver.exe')
context.driver = webdriver.Chrome(service=s, options=options,desired_capabilities=capa)
The --headless
option creates an error:
Timed out receiving message from renderer: 10.000
so I have commented that out.
Question: Why does my local always pass but my Jenkins is unstable even though they run tests on the same VM. How can I fix this?
Background
I have looked at this similar question but no solution from it. Unfortunately I have spent many hours trying to fix this.
Installed Package Versions (PIP)
Package Version
--------------------- ---------
allure-behave 2.9.45
allure-python-commons 2.9.45
async-generator 1.10
attrs 21.4.0
behave 1.2.6
cairocffi 1.3.0
CairoSVG 2.5.2
certifi 2022.6.15
cffi 1.15.1
charset-normalizer 2.1.0
chromedriver 2.24.1
click 8.1.3
colorama 0.4.5
cryptography 37.0.4
cssselect2 0.6.0
defusedxml 0.7.1
et-xmlfile 1.1.0
h11 0.13.0
idna 3.3
lxml 4.9.1
multipledispatch 0.6.0
numpy 1.23.1
openpyxl 3.0.10
outcome 1.2.0
pandas 1.4.3
parse 1.19.0
parse-type 0.6.0
Pillow 9.2.0
pip 22.0.4
pluggy 1.0.0
pycparser 2.21
pygal 3.0.0
pyOpenSSL 22.0.0
pypiwin32 223
pyshadow 0.0.4
PySocks 1.7.1
python-dateutil 2.8.2
python-docx 0.8.11
python-dotenv 0.20.0
pytz 2022.1
pywin32 304
requests 2.28.1
selenium 4.3.0
setuptools 58.1.0
six 1.16.0
sniffio 1.2.0
sortedcontainers 2.4.0
tinycss2 1.1.1
trio 0.21.0
trio-websocket 0.9.2
urllib3 1.26.10
webdriver-manager 3.8.1
webencodings 0.5.1
wsproto 1.1.0
Update 08/02/2022 4:11pm
Adding more implicit and explicit wait time fixed many of the flaky tests
Still facing some flaky tests only through Jenkins involving WebDriverWait(context.driver, 180).until(EC.presence_of_element_located((By.CSS_SELECTOR, '{$path}')))
Update 08/03/2022 8:20am SOLUTION
One issue was I needed to wait until an element becomes visible instead of just in the DOM. I had to change some presencee_of_element_located() to visibility_of_element_located():
Example
Before
WebDriverWait(context.driver, 10).until(EC.presence_of_element_located((By.CSS_SELECTOR, '{$path}')))
After
WebDriverWait(context.driver, 60).until(EC.visibility_of_element_located((By.CSS_SELECTOR, '{$path}')))
Summary
Using the correct expected condition is important. I even had to alter one of mine to element_to_be_clickable to make it not flaky. It was a test involving waiting for a clickable element to pop up anyway. Make sure to read the docs and use the best fit expected condition to avoid errors. Also, look into extending implicit or explicit wait times if its desync errors. Screenshotting upon fail can help you debug this.
There are couple of things you need to address as follows:
Presumably you are on a windows system, so you need to drop the argument:
options.add_argument("--disable-gpu")
as the argument --disable-gpu
was effective only for non-windows environment and is no more relevant.
Some of the arguments are no more relevant, so you can drop them:
options.add_argument("enable-automation")
options.add_argument("--disable-extensions")
Ideally jenkins test should be executed as a non-root / non-admin user, so you may not need the arguments:
options.add_argument('--no-sandbox')
options.add_argument("--disable-dev-shm-usage")
Possibly you don't require the either of the following arguments as well:
options.add_argument("--dns-prefetch-disable")
options.add_argument("--disable-browser-side-navigation")
options.add_argument("--force-device-scale-factor=1")
You may like to tweak the other argument as:
prefs = {"download.default_directory" : r'C:\Users\MyUser\Path\Directory\'}
options.add_experimental_option("prefs", prefs)
Finally, you aren't using any capabilities
so you can remove the argument from the constructor:
desired_capabilities=capa
Your effective code block can be:
options = Options()
options.add_argument("--headless")
options.add_argument("--window-size=1920,1080")
options.add_experimental_option('excludeSwitches', ['enable-logging'])
options.add_experimental_option("excludeSwitches", ["enable-automation"])
prefs = {"download.default_directory" : r'C:\Users\MyUser\Path\Directory\'}
options.add_experimental_option("prefs", prefs)
s = Service('chromeDriverPath/chromedriver.exe')
context.driver = webdriver.Chrome(service=s, options=options)