Can no longer Log In to a website I used to monitor with Playwright

[kultigspritzig]

I monitor a website for changes intermittently. Sometime between now and 3 months ago, a regression in the project or Playwright made it impossible to log in to that website. I am getting a 30s timeout.

URL: https://portalpacjenta.luxmed.pl/PatientPortal/NewPortal/Page/Account/Login
Steps:

• Enter text in field: #Login (any text)
• Enter text in field: #Password (any password)
• Click element: #LoginSubmit

Used to work just fine, now it just stalls with timeout. Is there any way to debug what's happening behind? I tried going back to 0.44.2 and using 1.60-chrome-stable tag for Playwright, to no avail.

[dgtlmoon]

heya

you dont say how you installed, or what operating system..

hmm so if you went back to 0.44.2 and it still didnt solve it, then it's not a "regression"

can you ZIP and attach the playwright/browserless logs? (please dont paste them as a huge comment)

and zip and attach the container logs?

[kultigspritzig]

you dont say how you installed, or what operating system..

Docker on Alpine 3.18.3

hmm so if you went back to 0.44.2 and it still didnt solve it, then it's not a "regression"

I didn't mean that it used to work with 0.44.2, I simply tried the previous stable version. I haven't used that particular watch in 3 months, so can't tell which was the version it last worked with. In any case, it definitely did work just fine, which is why I insist it is a regression.

For the record, I just tried with Press Enter instead clicking on #LoginSubmit and it worked. I also tried Click element containing text, but that didn't work. That having said, any subsequent Click element actions after logging-in also fail

can you ZIP and attach the playwright/browserless logs?

Sure can, but have you tried on your end and can confirm it works for you?

[dgtlmoon]

Sorry to be contrary here, but this doesnt make sense

I didn't mean that it used to work with 0.44.2

"by D Nir · 2008 · Cited by 29 — A regression bug is a bug which causes a feature that worked correctly to stop working after a certain event (system upgrade, system patching, etc)"

so can you move back to a version where it worked 100%? otherwise its not possible to prove it either way

[dgtlmoon]

The very definition of 'regression bug' means you can move back to whatever version and it should start working again, please, can you try that and let me know which version works perfectly?

[dgtlmoon]

I am getting a 30s timeout.

hmmm atleast on our paid subscriptions i can see it working

so yeah needs more log data from you i think

does it work "sometimes" or "never' ?

[dgtlmoon]

also be sure to use browserless/chrome:1.60-chrome-stable in your docker-compose

[dgtlmoon]

There is something weird in the playwright library

setting force=True seems to help but not solve it totally..

def action_enter_text_in_field(self, selector, value):
   self.page.fill(selector, value, timeout=15 * 1000, force=True)

even when all of these are correct according to https://playwright.dev/python/docs/actionability for .fill

def action_enter_text_in_field(self, selector, value):
    loc = self.page.locator(selector)
    print ("[is editable]", loc.is_editable())
    print ("[is enabled]", loc.is_enabled())
    print ("[is disabled]", loc.is_disabled())
    print ("[is visible]", loc.is_visible())
    self.page.fill(selector, value, timeout=15 * 1000, force=True)

output

[is editable] True
[is enabled] True
[is disabled] False
[is visible] True
[enter text in field] Attempting .fill of  INPUT[name="ctl00$content$login$UserName"]
Timeout 15000ms exceeded.
=========================== logs ===========================
waiting for locator("INPUT[name=\"ctl00$content$login$UserName\"]")
  locator resolved to <input type="text" class="login_fields" id="ctl00_conte…/>
elementHandle.fill("xxxxxxxxxxxxxx")
  waiting for element to be visible, enabled and editable
    forcing action
============================================================

I guess forcing action is not an exception there? unsure..

removing the timeout (set timeout=0) seems to help further slightly, but it means it never 'waits' for the element to be on the page