846402 - [Cost control] Cost control does not change values until the date is changed after flashing

Status: VERIFIED FIXED

(Firefox OS Graveyard :: Gaia::Cost Control, defect)

Description

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

## Environment :
Gecko  http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/3867fca22e3d
Gaia   a41c7921fbf0839b894f277261dad4a22507f979
BuildID 20130228070201
Version 18.0
Unagi

1) Flash the device
2) go through FTE without setting up wifi/mobile data
3) launch cost control
4) go through cost control setup
5) turn wifi/mobile data on
6) play on the net.
7) check cost control app again

Expected: numbers grow
Actual:
Numbers are 0;  It won't change the numbers until you move it to the next day.

Comment 1

[Dietrich Ayala (:dietrich)]

Daniel, should this block 1.0.1? Users might be spending money and it would not be clear because of this bug.

Comment 2

[Daniel Coloma:dcoloma]

Changing the ni to Salva

Comment 3

[Salvador de la Puente González [:salva]]

Sorry, but I'm not able to reproduce this. My numbers are increasing. Naoki, do not you see the numbers increasing, or the charts?

Comment 4

[Alex Keybl [:akeybl]]

Let's renominate if reproduced.

Comment 5

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

You have to flash the device.  It's an initialization issue.

This still occurs:

Master build : 2013-04-04-03-09-55
name="mozilla-central" revision="c232bec6974d"
name="integration/gaia-central" revision="d50316e8e9b9"
name="gecko.git" revision="4d9b24f0cb7cabfa1ceb4d2953e993e6af3ca1c4"
name="gaia.git" revision="29660ec0a4ad7c90d22e33852e1601d45f781e03"

Gecko  http://hg.mozilla.org/releases/mozilla-b2g18/rev/da523063aa7b
Gaia   a845be046c5d3cb077e3c78f963ca5c079e7ab3d
BuildID 20130404070202
Version 18.0
Unagi

Comment 6

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: v1.1 w/ overused pay-per-go sim card

Comment 7

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: master w/ monthly still in affect SIM card

Both screenshots are after playing around for a while using wifi.

Comment 8

[Salvador de la Puente González [:salva]]

Can we reproduce this, Carlos?

Comment 9

[Salvador de la Puente González [:salva]]

Naoki, can you try to cat the contents of /proc/net/dev in the device:

1- adb shell
2- cat /proc/net/dev

Please, do this before launching Cost Control at all; after passing the configuration for CC without using Internet and after surfing the net.

Let's discard if it is a back-end issue by seeing if there are data for interface wlan0

Comment 10

[Albert [:albert]]

Naoki, which is the timestamp set to the device after flashing? and after the FTU? What is the timezone set during the FTU?

Comment 11

[Tony Chung [:tchung]]

Leaving this on leo? nom for another day; for qa to try and reproduce from the new comments above.   If yes, we can nominate it.

Comment 13

[Joe Cheng [:jcheng] (please needinfo)]

(In reply to Tony Chung [:tchung] from comment #11)
> Leaving this on leo? nom for another day; for qa to try and reproduce from
> the new comments above.   If yes, we can nominate it.

adding QAWANTED

Comment 14

[Carlos Martínez Toral [:carlosmartinez]]

I´ve been able to reproduce the issue. Using v1.0.1 (Gecko-fe6592b.Gaia-b4d917d) in an unagi device with QC RIL (AU_LINUX_GECKO_ICS_STRAWBERRY_V1.01.00.01.19.070). 

STR:
1-Flash the device and install QC RIL
2-Go through FTE without configure anything (don´t activate 3G nor connect to a wifi, don´t change anything in TZ / date / hour screen)
3-Open cost control app and configure the app for your SIM
4-Go to settings and activate 3G
5-Surf the internet
6-Open cost control app

Expected result --> 3G data used should be tracked and shown in the widget and graph

Actual result --> 0 KB is shown in both the widget and graph

Comment 15

[Daniel Coloma:dcoloma]

Salva, can you take this one?

Comment 16

[Salvador de la Puente González [:salva]]

Hi there! It is a back-end problem. Albert, can you provide some extra info?

Comment 17

[Albert [:albert]]

(In reply to Salvador de la Puente González [:salva] from comment #16)
> Hi there! It is a back-end problem. Albert, can you provide some extra info?

Is a back-end problem, I'm working on it.

Comment 19

[Albert [:albert]]

This bug was solved in Bug 844774. The problem is that it wasn't landed for b2g18_v1_0_1, only for b2g18.

Comment 20

[Salvador de la Puente González [:salva]]

(In reply to Albert from comment #19)
> This bug was solved in Bug 844774. The problem is that it wasn't landed for
> b2g18_v1_0_1, only for b2g18.

Ok, so, reopening bug 844774 and adding it as a blocker. I renominated it tef? for be considered in the triage. See bug 844774 for further details.

Comment 22

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

This has not been resolved, so I am reopening this bug.  It is not a duplicate.

Comment 23

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: cat of proc net dev

results in attachment.

Comment 24

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: screenshot of the same device of the proc net dev

This is using a build:
Gecko  http://hg.mozilla.org/releases/mozilla-b2g18/rev/cfef36c0c8bc
Gaia   a80be95f553de517e5d8a159e04511cda5e38be4
BuildID 20130409070205
Version 18.0
Unagi

Comment 25

[Albert [:albert]]

It can be reproduced when timezone is set in the FTU to a 'minor' timezone than the pc used to flash the device. When it is set to the same or greater one it works properly.

Comment 26

[Albert [:albert]]

Gecko returns data usage, but the app doesn't show it:

I/Gecko   (  419): -*- NetworkStatsManager: result: {"connectionType":"wifi","start":"2013-04-10T04:00:00.000Z","end":"2013-04-30T04:00:00.000Z","data":[{"rxBytes":1168,"txBytes":44257,"date":"2013-04-10T00:00:00.000Z"},{"date":"2013-04-11T00:00:00.000Z"},{"date":"2013-04-12T00:00:00.000Z"},{"date":"2013-04-13T00:00:00.000Z"},{"date":"2013-04-14T00:00:00.000Z"},{"date":"2013-04-15T00:00:00.000Z"},{"date":"2013-04-16T00:00:00.000Z"},{"date":"2013-04-17T00:00:00.000Z"},{"date":"2013-04-18T00:00:00.000Z"},{"date":"2013-04-19T00:00:00.000Z"},{"date":"2013-04-20T00:00:00.000Z"},{"date":"2013-04-21T00:00:00.000Z"},{"date":"2013-04-22T00:00:00.000Z"},{"date":"2013-04-23T00:00:00.000Z"},{"date":"2013-04-24T00:00:00.000Z"},{"date":"2013-04-25T00:00:00.000Z"},{"date":"2013-04-26T00:00:00.000Z"},{"date":"2013-04-27T00:00:00.000Z"},{"date":"2013-04-28T00:00:00.000Z"},{"date":"2013-04-29T00:00:00.000Z"},{"date":"2013-04-30T00:00:00.000Z"}]}

I/Gecko   (  419): -*- NetworkStatsManager: result: {"connectionType":"mobile","start":"2013-04-10T04:00:00.000Z","end":"2013-04-30T04:00:00.000Z","data":[{"rxBytes":28464,"txBytes":314443,"date":"2013-04-10T00:00:00.000Z"},{"date":"2013-04-11T00:00:00.000Z"},{"date":"2013-04-12T00:00:00.000Z"},{"date":"2013-04-13T00:00:00.000Z"},{"date":"2013-04-14T00:00:00.000Z"},{"date":"2013-04-15T00:00:00.000Z"},{"date":"2013-04-16T00:00:00.000Z"},{"date":"2013-04-17T00:00:00.000Z"},{"date":"2013-04-18T00:00:00.000Z"},{"date":"2013-04-19T00:00:00.000Z"},{"date":"2013-04-20T00:00:00.000Z"},{"date":"2013-04-21T00:00:00.000Z"},{"date":"2013-04-22T00:00:00.000Z"},{"date":"2013-04-23T00:00:00.000Z"},{"date":"2013-04-24T00:00:00.000Z"},{"date":"2013-04-25T00:00:00.000Z"},{"date":"2013-04-26T00:00:00.000Z"},{"date":"2013-04-27T00:00:00.000Z"},{"date":"2013-04-28T00:00:00.000Z"},{"date":"2013-04-29T00:00:00.000Z"},{"date":"2013-04-30T00:00:00.000Z"}]}

Please Salva can you check cost control app?

Comment 27

[Salvador de la Puente González [:salva]]

Ok, problem located. I'm going to explain what was happening here:

To support SIM switching scenarios there is a component inside Usage that fixes samples returned by the NetworkStats API. Let's say this component has to compare two dates to check if they refer to the same date (discarding time). If not, the usage for that SIM, that day, is discarded.

One of the dates, say Date A, is set when enabling the Usage application (we have very similar working times so we can say we do this between 09:00 AM - 06:00 PM, local time). The other, Date B, is always the present day at 00:00 UTC.

Check these situations (24h format):
------------------------------------

Madrid (GTM+2):
         +==================+==================+
         |    LOCAL TIME    |        UTC       |
         +==================+==================+
  Date A | 2013-04-10 09:00 | 2013-04-10 07:00 |
         +------------------+------------------+
  Date B | 2013-04-10 02:00 | 2013-04-10 00:00 |
         +------------------+------------------+

So both refer to the same local date 2013-04-10

San Francisco (GMT-7):
         +==================+==================+
         |    LOCAL TIME    |        UTC       |
         +==================+==================+
  Date A | 2013-04-10 09:00 | 2013-04-10 16:00 |
         +------------------+------------------+
  Date B | 2013-03-10 17:00 | 2013-04-10 00:00 |
         +------------------+------------------+

So, they refer to different local dates.

Of course, the problem is taking in count local times instead of UTC times.

Comment 28

[Salvador de la Puente González [:salva]]

STR:

1. Make reset-gaia
2. In FTU, set a timezone with the form GMT - XX (GMT minus something) such as New York. 
3. Configure Usage
4. Browse the Internet
5. Check Usage

Expected:
Data usage counters increase.

Actual:
Data usage counters remain 0.

Comment 29

[Salvador de la Puente González [:salva]]

Attachment: Use UTC comparison instead of local time to avoid the effect of timezones.

Comment 30

[Francisco Jordano [:arcturus] [:francisco]]

Comment on attachment 735835 [details]
Use UTC comparison instead of local time to avoid the effect of timezones.

Pretty clean solution :)

Nice work as always!

Comment 31

[Salvador de la Puente González [:salva]]

Master: 9c0320c1787760af16e76b79932b25546c1f8a9f

Comment 32

[John Ford [:jhford] CET/CEST Berlin Time]

Uplifted commit 9c0320c1787760af16e76b79932b25546c1f8a9f as:
v1-train: ea54fe7edc001152842a6193652da52984f7a31d
v1.0.1: ae65b2bae3bb7c4e70a99a138618365e43fe64fd

Comment 33

[Sarah Parsons]

Verified fixed on Unagi Build ID: 20130430070204

Environmental  Variables:
Kernel Date: Dec 5
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/8ca8149ce236
Gaia: b525c25063d33d2e073d9f3e1e1164fadefec998

Also verified fixed on Inari Build ID: 20130430070201

Environmental  Variables:
Kernel Date: Feb 21
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/010498e599a7
Gaia: de0f1fc6c58616b8a33fca482f1f8684d4e74e9e

Usage is tracked and displays properly.