614112 - Push Firefox Input 2.0.1 to production

Status: VERIFIED FIXED

(Infrastructure & Operations Graveyard :: WebOps: Other, task)

Description

[Fred Wenzel [:wenzel]]

Two annoying regressions popped up on Input after the push to 2.0 last week. Bug fixes are on the 2.1 development branch, but since the bugs are obvious to users, we need to push these fixes, backported to 2.0.

We've tagged the prod branch, so running:
/root/bin/input_update.sh 

is all that needs to be done for this push.

Downtime: no.
User impact: restoration of expected behavior.

Comment 1

[Fred Wenzel [:wenzel]]

Oremj, can you help us with this? As there are unsettling amounts of tracebacks coming from production, we'd like to push this asap, at the latest tomorrow.

Comment 2

[Phong Tran [:phong]]

Are you sure we don't need downtime for this?

Comment 3

[Fred Wenzel [:wenzel]]

Neither database nor Sphinx changes, so no downtime. However, if you want to announce it and push during tomorrow's push window, go ahead. Better safe than sorry.

Comment 4

[Shyam Mani [:fox2mike]]

Good to go at 4?

Comment 5

[Shyam Mani [:fox2mike]]

Grabbing it assuming it's at 4. I'll wake up for it.

Comment 6

[Fred Wenzel [:wenzel]]

Yes!

Comment 7

[Fred Wenzel [:wenzel]]

Over in bug 611109, it turned out that the prod database possibly has tables and/or fields that are not UTF-8. Please, as part of the push, go in and change these to UTF-8. This will also fix such issues we encountered on Staging when the prod database is copied over the next time.

Comment 8

[Shyam Mani [:fox2mike]]

[root@mradm02 ~]# /root/bin/input_update.sh 
Updating vendor...
Already up-to-date.
Updating reporter...
remote: Counting objects: 95, done.
remote: Compressing objects: 100% (69/69), done.
remote: Total 70 (delta 52), reused 0 (delta 0)
Unpacking objects: 100% (70/70), done.
From git://github.com/fwenzel/reporter
 * [new branch]      last_known_good -> origin/last_known_good
   7e10b33..fbd440a  master     -> origin/master
 * [new branch]      mobile-deviceinfo-610494 -> origin/mobile-deviceinfo-610494
   61a3aea..5aed9c8  prod       -> origin/prod
 * [new branch]      urlparse-traceback-614020 -> origin/urlparse-traceback-614020
 * [new tag]         2.0.1      -> 2.0.1
M       locale
Previous HEAD position was 61a3aea... L10n update (automatic commit)
HEAD is now at 5aed9c8... 2.0.1 reverts
remote: Counting objects: 20, done.
remote: Compressing objects: 100% (12/12), done.
remote: Total 15 (delta 7), reused 0 (delta 0)
Unpacking objects: 100% (15/15), done.
From git://github.com/fwenzel/reporter-locales
   8ad6589..de7baac  master     -> origin/master
Submodule path 'locale': checked out 'de7baac6132f5bc52d9641a1b9ac7491d5cece29'

Should be good.

Checking the DB stuff now.

Comment 9

[Shyam Mani [:fox2mike]]

Modified the following tables to utf8 from latin1 :

celery_taskmeta
celery_tasksetmeta
theme

They're all MyISAM still. Since the theme table had no specific charset definitions for it's fields, changing the table to utf8 should suffice.

Comment 10

[Stephen Donner [:stephend] Not actively reading bugmail]

The traceback and pagination bugs we shipped in this release have been verified; http://input.mozilla.com/en-US/search/?q=&product=firefox&version=4.0b7&date_start=2010-11-22&date_end=&sentiment=happy, however, it still showing messages from 55+ minutes ago (caching?), so I'd like to wait a bit to verify this release.

The latest feedback does seem to be pouring in on all platforms just fine (on the homepage, "Latest Feedback" block), and I tested submission both on mobile and desktop sites, but it's so data-driven that I can't see everything show up just yet :-(

Comment 11

[Stephen Donner [:stephend] Not actively reading bugmail]

Verified FIXED; post-push, I filed bug 614484 to cover comment 10, but am seeing it looking much, much better (without code changes); I'm keeping a wary eye on it, but looks good.