1736614 - regression range incorrectly narrowed to later nightly on same day

Status: UNCONFIRMED

(Testing :: mozregression, defect, P3)

Description

[Kevin Locke]

Attachment: mozregression --debug --good 2021-10-07 --bad 2021-10-09 with main 7e6b2c5

With mozregression 4.0.18 or main 7e6b2c5, when given dates for good/bad, mozregression launches the first nightly for the given date then uses the changeset id from the second nightly as the beginning of its range. For example, with mozregression --good 2021-10-07 --bad 2021-10-09 it first launches build 20211007091917 changeset 796cb80eb626f762dfac0290f62744888b62de77 to confirm good, then build 20211009212622 changeset 658c01aa35228543db3728393f7b2bfa626249df to confirm bad, then proceeds with Last good revision: 3cf700ab977a298e19a00f1ec66d14939766955b (which is the changeset id of build 20211007215152 which was not tested, and in my case was actually bad, leading to an incorrect range).

For reference, I was also able to reproduce the issue using mozregression 4.0.0. I could test with older releases if that is likely to be useful.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:wlach, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 2

[William Lachance (:wlach)]

That's unfortunate. It's likely that this problem has existed for a while now (maybe since we started producing 2 nightlies/day), the logic for choosing builds hasn't changed recently.

Unfortunately I'm not sure when I'll have time to look into this, but I think the above description should be enough to start a debugging session and see what's going on.