1568789 - TEST-UNEXPECTED-FAIL | [snip]/folder-display/test-selection.js | test-selection.js::test_enter_scroll_to_new

Status: RESOLVED WORKSFORME

(Thunderbird :: General, defect)

Description

[Jorg K (CEST = GMT+2)]

TEST-UNEXPECTED-FAIL | /builds/worker/workspace/build/tests/mozmill/folder-display/test-selection.js | test-selection.js::test_enter_scroll_to_new has started to pop up first on my try here:

https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=011167f1ac2a0d861f9514df9ef437347f0815f7

I did that try run since I had nothing to push to cover the M-C merge of
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=5585edba8fdb83267722e62241952272c8e24a8f&tochange=6598e37c88d2816deed4fdaedbddf9c9dade7987

As one can see, I reran some tests, and the Z2's turned green. Also this push
https://treeherder.mozilla.org/#/jobs?repo=comm-central&revision=e2dad5423b3ba2aa4742d4f0e74e0a06ebe92291
only has one of them.

They are back here now:
https://treeherder.mozilla.org/#/jobs?repo=comm-central&selectedJob=258265997&revision=33aa95186ee7d2c81dc6d17c44160b10db97ecbd
but not consistently.

Log says:
https://taskcluster-artifacts.net/DmIFlCsSQ5-ku7prwqu1IQ/0/public/logs/live_backing.log
[task 2019-07-25T10:07:20.574Z] 10:07:20 INFO - EXCEPTION: View index 19 is not visible! (0-12 are visible)

They pass locally for me:
mozmake SOLO_TEST=folder-display/test-selection.js mozmill-one

Maybe the front-end cracks want to take a look.

Comment 1

[Jorg K (CEST = GMT+2)]

Very patchy failure, last push has one such failure:
https://treeherder.mozilla.org/#/jobs?repo=comm-central&revision=047998237781d0e3c85c90e798bdba36e8bd7d92

Comment 2

[Alessandro Castellani [:aleca]]

I'm randomly stumbling upon 4 failures locally on Linux, as they happen if I trigger the entire folder instead of individual tests.

This is the first:

EXCEPTION: messages box height (659.6666870117188) not equal to the sum of
displayDeck height (0) and message pane box wrapper height (631.3333740234375)
and message notification box height (28.333328247070312):
'659.6666870117188' != '659.6667022705078'.

And these are the other 3:
EXCEPTION: Thought we would find row 1 at 40,27 but we found -1

I'll see if I can fix them.

Comment 3

[Jorg K (CEST = GMT+2)]

I'm getting old :-( - This was already reported in bug 1528482. So it's not new at all.

Comment 5

[Jorg K (CEST = GMT+2)]

This is almost perma-red now. Can we get some tracking on this?

Comment 6

[Geoff Lankow (:darktrojan)]

I've been trying to fix this when I have a few spare moments, but Mac OS is defying me at the moment. My fix which seems to be good for the other platforms makes Mac fail every time.

Comment 7

[Geoff Lankow (:darktrojan)]

And in fact there's another failure now. test_selection_last_message_deleted is also failing sometimes.

Comment 8

[Alessandro Castellani [:aleca]]

On Linux, I can recreate the failures only if I run the entire folder:
make -C obj-x86_64-pc-linux-gnu mozmill-one SOLO_TEST=folder-display/

Running the single tests don't cause the failures.
This is gonna be interesting to debug.

Comment 9

[:aceman]

That does happen. In mozmill all the tests in the folder are run in the same TB session so whatever is left after one test the next one will see. Also, tests aren't always run in the same order. So tests have to be neutral enough to not leave garbage after them.

So if you can't reproduce the failure with single test, see in what state your test actually starts after the others in the folder are run.
Also you can get screenshots on the server showing in what state the test has failed. Then you can see if there are some leftover folders or messages from previous tests.

Comment 10

[Geoff Lankow (:darktrojan)]

I found I could make it fail every time (even if only test_enter_scroll_to_new is run) by inserting mc.sleep(1000) before the assert_visible. There's something wrong with the timing, which does go away by waiting after the first enter_folder and after make_new_sets_in_folder, but not on Mac for some reason.

Comment 11

[Geoff Lankow (:darktrojan)]

See this Try run for my most recent attempt and a screenshot at the point of failure. (No idea why the folder tree is transparent!)

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/comm-central/rev/f07021063440
Temporarily disable two subtests of MozMill test folder-display/test-selection.js on Linux and Mac. rs=bustage-fix

Comment 13

[alta88]

I bet if you move these two to be last, it will fix this.

test_selection_select_column()
test_selection_select_column_deselection()

(Only test_selection_last_message_deleted fail, the other predates).

Comment 14

[Jorg K (CEST = GMT+2)]

What's the bet? ;-)
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=bd1324f20336d8a0962999adbb5bd79a3fb051d2

Comment 15

[Jorg K (CEST = GMT+2)]

Attachment: 1568789-move-some-test-last.patch

Looking good so far. So why do these tests upset the rest?

Comment 16

[Jorg K (CEST = GMT+2)]

https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=bd1324f20336d8a0962999adbb5bd79a3fb051d2&selectedJob=263309149
TEST-UNEXPECTED-FAIL | /builds/worker/workspace/build/tests/mozmill/folder-display/test-selection.js | test-selection.js::test_enter_scroll_to_new

You lose. What was the bet? Do I get a beer now? ;-)

Comment 17

[alta88]

Nope, the bet is that test_selection_last_message_deleted (comment 7) no longer fails (comment 13). The second of the column tests deleted a row and maybe moved the scroll view, that broke the very fragile _row_click_helper(). The scroll test failed before. So pay up. ;)

Comment 18

[Jorg K (CEST = GMT+2)]

Well, comment #13 was a little vague and you edited it to add that clause. But I'm happy to pay. What did we agree on? ;-)

Comment 19

[Intermittent Failures Robot]

1 failures in 5144 pushes (0.0 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• comm-central: 1

Platform breakdown:

• windows10-64: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1568789&startday=2019-12-02&endday=2019-12-08&tree=all

Comment 20

[Intermittent Failures Robot]

6 failures in 4867 pushes (0.001 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• comm-central: 6

Platform breakdown:

• windows7-32: 3
• windows10-64: 2
• windows10-64-shippable: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1568789&startday=2019-12-09&endday=2019-12-15&tree=all

Comment 21

[Intermittent Failures Robot]

2 failures in 4582 pushes (0.0 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• comm-central: 2

Platform breakdown:

• windows10-64: 1
• windows7-32: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1568789&startday=2019-12-16&endday=2019-12-22&tree=all

Comment 22

[Wayne Mery (:wsmwk)]

Where to go from here?
No intermittent reports since 2020.

Comment 23

[Magnus Melin [:mkmelin]]

All intermittent that didn't happen in a log time can be closed. If they re-emerge, something regressed and cause would be different.
The tests are re-enabled these days.