End2End testing. Issue with React job
kuvakina opened this issue · 6 comments
As we tried to see a pattern in End2End testing for React, we noticed some issues:
- Until Monday, 16.07.2018, jobs on React were failures. The fails were the same as for Angular before Monday #193
And also the manual test showed that tests work correctly and the problem with the response from server:
- Allergies failure:
This is the failure in the report:
We try to add allergy:
Response from the server that changes were saved:
And we don't see changes:
The same was for PersonalNotes, Contacts, and Vaccination.
There were more or fewer failures, it was different from test to test.
- After Tony reboot EtherCIS, we were left with two failures, that we couldn't reproduce:
- patientAccordionTest:
This is the failure in the report:
This is what we see:
-patientInformationTest:
This is the failure in the report:
And what we see:
- As we tried to run tests on the Remote desktop today, we got the same failure with patientAccordionTest:
But we couldn't continue the test after this, because we were thrown out from Remote desktop shortly after the next test started:
- And the most confusing part is that local tests run mostly just fine. So we can't say, that the problem either with tests or code.
We'd like to ask maybe @robtweed help.
UPD [10.08.2018]:
On Jenkins fails the test on React:
<
✕ patientInformationTest
Patient Information page
✓
Element <input[name^='username']> was visible after 2336 milliseconds.
✓
Element <button.auth0-lock-submit> was not present after 3291 milliseconds.
✕
Timed out while waiting for element <Section[name=preferences],Element[name=@title]> to be present for 25000 milliseconds. Expected "visible" but got: "not found"
/>
And for now, we think it could be because of:
-
Tests memory is used to the max and the test just can't complete the operation.
-
It could happen that login to securedev.ripple.foundation loads for too long, and that makes the test to fail.
-
There is process on a remote desktop (when we use VNC), and it also took a memory, that makes the test fail.
If it's the third variant, that it means I could not kill the process yesterday when I checked tests via VNC. I'll check this on weekend and write a feedback on Monday.
Last week tests for React failed.
The problem was with a single test:
We think it was because at some point test was running out of memory.
At the moment tests run successfully and don't fail: https://www.screencast.com/t/RXPzZmmp
But we'll monitor them for now.
We can monitor the memory using htop Command Under SSH Terminal.
Once your server is connected with SSH, you can run Htop command by just typing htop and press Enter.
A screen will open up as a result of this command where you can see where jobs run and what memory they use:
As I said earlier, the tests run fine now, but I'll continue monitoring them.
Also, we'll discuss with Kirill connection of this issue with that one: #130
Since last week (last week of September), for some reason React test started to fail. The problem occurred with patientAccordionTest & patientInformationTest:
https://www.screencast.com/t/lnQt0JXM
Problem appeared as there wasn't enough time/memory to run tests.
While trying to connect to Remote machine today, we noticed malware on server (we fixed this already): https://www.screencast.com/t/LYZ7CndQ6K0o
So we can say, that tests run fine and on the other hand, if React test starts to fail - we had to check server.
But we don't know how it got infected.
Could we consider tasks with E2E testing Done?
#149
#130
#79
No further issues since this was resolved, closing issue