Add debug code to flaky field_data test #15535
Conversation
|
The current CI failure seems to be something else... jenkins, test this |
|
@Bargs did you give up on this PR or you still looking for a review for it? |
💔 Build Failed |
|
@stacey-gammon I was waiting for review, then forgot about it :) I dunno if this test has continued to be flaky in the intervening time or not |
Bargs
force-pushed
the
debugFieldDataTest
branch
from
9743230
to
a9b6b51
Compare
|
Ping @stacey-gammon @weltenwort I've rebased this PR, please take a look if you get a chance. |
💚 Build Succeeded |
There was a problem hiding this comment.
Could not reproduce the targeted failure locally, but this looks like it could give some insight.
Would be sweet if Kibana offered a hook for the tests to gather the request details. Alternatively, we could inject some JS via driver.executeScript to hook into the xhr requests ourselves or communicate with the devtools to save the HAR. The former would be more compatible across browsers, I assume.
Anyway, that's for a different PR 😉
|
did you confirm that a failure will spit out the debug info? I tried to simulate a failure, but I get:
instead of the debug output. But maybe something funky on my local. (btw if you run into other issues running tests locally, you might need to bump chromedriver temporarily). |
|
I did confirm it worked, but that was back in 2017 😆 I'll test it out and make sure. |
💔 Build Failed |
Bargs
force-pushed
the
debugFieldDataTest
branch
from
a9b6b51
to
1332fc5
Compare
|
@stacey-gammon so it looks like the UI changed since I first wrote this, I just had to update a helper to handle that change. You should be able to test the console output by simply changing the expected string now. |
💔 Build Failed |
|
Failing tests don't seem to be related to this change, retrying. Jenkins, test this |
💚 Build Succeeded |
Gathering more info for #13163
This PR re-enables a flaky test and adds some code to log additional debug info the next time it fails. As I mentioned in a comment in the linked issue, I suspect there's an actual bug in our response handling or _source formatting code. I dug through some of the code that I thought might be responsible but I didn't see any obvious issues. So before blindly searching any further I want to see the actual ES response for one of these failures and be 100% sure that the issue isn't with the data ES is sending back. If the ES data is fine I may need to figure out another way to zero in on the issue.