1321400 - LATEST_FIREFOX_VERSION is set to 45.5.1 instead of 50.0.2

Status: RESOLVED FIXED

(Release Engineering :: Applications: Shipit, defect)

Description

[Michal Stanke (Mozilla.cz) [:mstanke][:MikkCZ]]

Maybe during the hurry of releasing the 0day fix, there appears to be version 45.5.1 (corresponds to ESR) instead of 50.0.2 as the release version.

Comment 1

[Sylvestre Ledru [:Sylvestre]]

Seems fixed now
https://product-details.mozilla.org/1.0/firefox_versions.json
I wonder how we could get that. Maybe we should set a fail safe?!
Guys, any clue?

Comment 2

[Johan Lorenzo [:jlorenzo]]

I found some STRs:
1. On a dev env of ship it: create 50.0.2 and mark it as ready
2. With a mysql client, execute:
> UPDATE `shipit`.`firefox_release` 
> SET 
>     `complete` = TRUE,
>     `status` = 'shipped',
>     `shippedAt` = NOW()
> WHERE
>     `name` = 'Firefox-50.0.2-build1';

3. Repeat step 1 and 2 with 45.5.1esr
4. Go to http://localhost:5000/json/1.0/firefox_versions.json and see 45.5.1 being declared as the latest release.

Root cause: the way we call getFilteredReleases() allows stability releases to be returned[1]. 45.5.1 being a stability release which has a more recent timestamp makes it appear as the latest Firefox available.

I'm not too sure how to tackle the fix (without breaking /firefox_history_stability_releases.json). I'll give it a try.

[1] https://github.com/mozilla-releng/ship-it/blob/0cac23ab2bf379828b567ef769c2771ca2b287ec/kickoff/jsonexport.py#L127

Comment 3

[Johan Lorenzo [:jlorenzo]]

Attachment: PR

I have a fix that I'm not particularly proud of. It has some unwanted side effects (like breaking tests). Please don't look at it too closely, I'll revisit it :)

Comment 5

[Johan Lorenzo [:jlorenzo]]

Bug 1321411 comment 0 has a very detailed explanation about the root cause :)

Comment 6

[Liz Henry (:lizzard) (relman/hg->git project)]

This just happened again today with 45.6.0esr and 50.1. We normally release ESR just after the main release.

Comment 7

[Nick Thomas [:nthomas] (UTC+12)]

Fixed the db with:

mysql> select name, status, submittedAt from firefox_release where name="Firefox-45.6.0esr-build1" or name="Firefox-50.1.0-build1";
+--------------------------+---------+---------------------+
| name                     | status  | submittedAt         |
+--------------------------+---------+---------------------+
| Firefox-45.6.0esr-build1 | shipped | 2016-12-09 23:19:34 |
| Firefox-50.1.0-build1    | Started | 2016-12-08 01:58:56 |
+--------------------------+---------+---------------------+
2 rows in set (0.00 sec)

mysql> update firefox_release set submittedAt="2016-12-08 01:50:12" where name="Firefox-45.6.0esr-build1";
Query OK, 1 row affected (0.02 sec)
Rows matched: 1  Changed: 1  Warnings: 0

mysql> select name, status, submittedAt from firefox_release where name="Firefox-45.6.0esr-build1" or name="Firefox-50.1.0-build1";
+--------------------------+---------+---------------------+
| name                     | status  | submittedAt         |
+--------------------------+---------+---------------------+
| Firefox-45.6.0esr-build1 | shipped | 2016-12-08 01:50:12 |
| Firefox-50.1.0-build1    | Started | 2016-12-08 01:58:56 |
+--------------------------+---------+---------------------+
2 rows in set (0.01 sec)

Comment 8

[Rail Aliiev [:rail]]

I'll take it.

Comment 9

[Rail Aliiev [:rail]]

Attachment: PR #2

Another approach with version sort, so we can ship 47.0.2 after 49.0.

Comment 10

[Jordan Lund (:jlund)]

(In reply to Rail Aliiev [:rail] ⌚️ET from comment #9)
> Created attachment 8818613 [details] [review]
> PR #2
> 
> Another approach with version sort, so we can ship 47.0.2 after 49.0.

will review this tomorrow

Comment 12

[Rail Aliiev [:rail]]

Comment on attachment 8818613 [details] [review]
PR #2

To be deployed

Comment 13

[Rail Aliiev [:rail]]

In production.