888833 - [OTA] After tried to manually check for updates, it showed error message

Status: RESOLVED WONTFIX

(Firefox OS Graveyard :: Gaia::Settings, defect)

Description

[Walter Chen[:ypwalter][:wachen]]

Full Leo Run4 Build in Leo

STR:
 1. Turn on data connection (and data roaming if roaming) in "Settings" > "Cellular and Data"
 2. Go to "Settings" > "Device Information" > "Check Now"

Expected Result:
  It should check and respond if the device is up-to-date or tell us about if there are update packages available

Actual Result:
  "There was an error when checking for updates" message came out, but the OTA notification still came out.

Comment 1

[bhavana bajaj [:bajaj]]

needinfo on walter to get the build id here with which he is hitting this case.Sounds like a regression to us, but adding qawanted to confirm .

Comment 2

[Jason Smith [:jsmith]]

(In reply to bhavana bajaj [:bajaj] from comment #1)
> needinfo on walter to get the build id here with which he is hitting this
> case.Sounds like a regression to us, but adding qawanted to confirm .

I think I saw this as well when I testing some of the app update bugs. Note that this means that the updates are found, but the error message is wrong. The STR however given in comment 0 for me isn't consistent STR, however.

I'm pretty sure this is a regression.

Can't remember the exact build ID I saw this on, but it was on a recent v1-train build.

Comment 3

[Walter Chen[:ypwalter][:wachen]]

Leo Partner Build v08h
This it for leo run4

Comment 4

[Jason Smith [:jsmith]]

I saw this again with the following STR:

1. Flash the build for 7/7/2013 b2g18 on unagi 
2. Check for updates manually - see 2 updates available
3. When you are 100 KB in, cancel the download
4. Check for updates again

Result - you'll get the error message cited in this bug.

Let's get a regression window here to figure out when this first became broken.

Comment 5

[Lukas Blakk [:lsblakk] use ?needinfo]

Blocking until we know more but if this is an OTA regression then it's obviously a blocker.

Comment 6

[Natalya Kot [:nkot]]

(In reply to Jason Smith [:jsmith] from comment #4)
> I saw this again with the following STR:
> 
> 1. Flash the build for 7/7/2013 b2g18 on unagi 
> 2. Check for updates manually - see 2 updates available
> 3. When you are 100 KB in, cancel the download
> 4. Check for updates again
> 
> Result - you'll get the error message cited in this bug.
> 
> Let's get a regression window here to figure out when this first became
> broken.

regression-window:
2013-05-28-07-02-09 - does not repro
2013-05-28-23-02-07 - repros	
2013-05-29-07-02-08 - repros

Comment 7

[Jason Smith [:jsmith]]

bug 862785 might have caused this regression.

Kaze - Could bug 862785 cause this regression? Or do you think it's something else?

Comment 8

[Alex Keybl [:akeybl]]

dhylands, would be great to get your opinion hear as well, oh update guru

Comment 9

[Dave Hylands [:dhylands]]

I downloaded this build:
https://pvtbuilds.mozilla.org/pub/mozilla.org/b2g/nightly/mozilla-b2g18-unagi/2013/07/2013-07-07-07-02-16/

It told me that there were 2 updates available, one being a system update, and one being an app update of Marketplace.

After cancelling the download (I couldn't cancel at 100 Kb, the fastest I could cancel was after a Mb or so was downloaded).

Anyways, I saw the "There was an error when checking for updates", but it seems to be related to the Marketplace app. If I went to the notification bar, it told me that there was 1 update available (just the system update) and I was able to resume and download it fine.

After applying the system update, pressing Check Now showed "This is already the latest version of Firefox".

I'm not that familiar with the app update stuff, so I'm not sure why it would decide to error out.

Comment 10

[Jason Smith [:jsmith]]

Etienne - Any ideas why the app update here would error out?

Comment 11

[Etienne Segonzac (:etienne)]

(In reply to Jason Smith [:jsmith] from comment #10)
> Etienne - Any ideas why the app update here would error out?

I think we console.info the error name for app updates when we get one, do we have a logcat?

Comment 12

[Jason Smith [:jsmith]]

QA Wanted to reproduce this bug and attach a logcat

Comment 13

[Natalya Kot [:nkot]]

Attachment: logcat

updated from:

Build  20130712070210
Gecko  http://hg.mozilla.org/releases/mozilla-b2g18/rev/282b5c37cf8d
Gaia   e2ef782119b7e79fc62c48d36f0c36909d982988

Comment 14

[Jason Smith [:jsmith]]

Relevant logcat:

07-17 13:26:31.182: E/GeckoConsole(794): Content JS WARN at app://settings.gaiamobile.org/gaia_build_defer_index.js:3 in consoleWarn: [l10n] #active-update is undefined.
07-17 13:26:31.182: E/GeckoConsole(794): Content JS ERROR at app://settings.gaiamobile.org/js/about.js:134 in onUpdateStatus: Error checking for system update: active-update

Comment 15

[Arthur Chen [:arthurcc][inactive after 6/30]]

It seems like gaia does not handle the newly added state 'active-update'.

Comment 16

[Arthur Chen [:arthurcc][inactive after 6/30]]

Dave, since 'active-update' indicates that the update is downloading, can we simply consider it as 'check-complete' in gaia?

Comment 17

[Dave Hylands [:dhylands]]

I think we need 2 solutions.

The proper solution would be to treat this as a new condition and display an appropriate error string. This would require a localized string, and I'm not sure where we are as far as introducing new strings. The new string should probably indicate that there is an update download in progress.

If, strings are frozen, then we could probably treat it as check-complete, but I would file a followup bug to fix it properly.

Comment 18

[Dave Hylands [:dhylands]]

Actually, something seems wrong here, as the "active-update" status only happens when you cancel the download.

In that case, it should just not generate an error. If we got an "active-update" as part of doing a check for updates, then it should behave as I described in comment 17.

Comment 19

[Arthur Chen [:arthurcc][inactive after 6/30]]

It seems like there are two separate problems here and the STR are different. I can reproduce the one reported by Walter. The following is the status I got with the latest LEO V08h image and LEO V08j image (the latest image).

V08h: apps.updateStatus : check-complete
      gecko.updateStatus: check-error-2152398862
V08j: apps.updateStatus : check-complete
      gecko.updateStatus: check-error-http-404

The second one can be reproduced following the steps in comment 4. Then we get the 'active-update' status as comment 14. We might need to file another bug as suggested in comment 17.

Any ideas on the check-error status of the first problem?

Comment 20

[Tim Guan-tin Chien [:timdream] (please needinfo)]

(In reply to Arthur Chen [:arthurcc] from comment #19)
> Any ideas on the check-error status of the first problem?

Arthur, are you blocked because of you need the answer above to work on this bug?

Comment 21

[Arthur Chen [:arthurcc][inactive after 6/30]]

Yes, just want to make sure that there are two problems here. The reported one seems a gecko bug. And the one reported in comment is a gaia bug. If this is true, I'll un-assign this bug from myself and file another gaia one.

Comment 22

[Arthur Chen [:arthurcc][inactive after 6/30]]

schien, any idea on comment 19?

Comment 23

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

check-error-2152398862 is corresponding to NS_ERROR_NET_TIMEOUT, which means packets are not delivered in time. This error could be cause by network congestion, weak Wi-Fi signal, etc. IMO we need to have a default string for unhandled error status.

Comment 24

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

Attachment: tcpdump of accessing leo update server

I capture a tcpdump of accessing leo update server (165.244.60.79) from desktop manually. There will be some duplicate/out-of-sync tcp packets, I think this is the reason of check-error-2152398862.
BTW, the update URL in leo points to a 404 page. We might need to tell our partner to setup a update.xml with no updates if we want to show "no update" message while manual check.

Comment 25

[Arthur Chen [:arthurcc][inactive after 6/30]]

schien, thanks for the information. For check-error-2152398862, is that a gecko bug?
Leo, could you help check if update.xml is set correctly?

Comment 26

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

(In reply to Arthur Chen [:arthurcc] from comment #25)
> schien, thanks for the information. For check-error-2152398862, is that a
> gecko bug?
I think it is a defect in the server side, will need leo to check the network/router for their update server.

Comment 27

[pupplerain78]

schien, I found the update url is wrong in your attacted tcpdump.
How did you get the leo image?

Comment 28

[Arthur Chen [:arthurcc][inactive after 6/30]]

Per comment 27, needinfo schien.

Comment 29

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

The leo image is D300_V08m_00_JUL_21_TLF_ES.zip

Comment 31

[Preeti Raghunath(:Preeti)]

:schien,

Are you working on this bug from dev?

Comment 32

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

No, based on comment 26 and comment 27, the update URL in leo image gave by partner is not pointing to a correct update server. We'll need partner to double check their setting in build system and update server.

Comment 33

[Alex Keybl [:akeybl]]

Wayne, can you follow up with Leo?

Comment 34

[Wayne Chang [:wchang]]

Leo is asking their FOTA team to check and clarify here.

Did we make any changes at all after flashing whats mentioned in comment 29? 

pupplerain78 - can you comment here what you suspect and/or need us to check?

Comment 35

[pupplerain78]

in case of buildid-20130614150813,
it is different from url scheme in leo release.
the update url in leo image is pointing a correct update server. but tcpdump on comment 24 does not.
did you modify "app.update.url"preperence by any chance?

Error-2152398862 could be occured because the FOTA server scheme has been applied incorrectly.
the issue associated with the server url does not have to take care.

Comment 36

[Arthur Chen [:arthurcc][inactive after 6/30]]

Are you suggesting that the current server scheme is incorrect?

Comment 37

[Arthur Chen [:arthurcc][inactive after 6/30]]

As comment 19, the first problem will be solved by the partner. Mark the bug as WONTFIX. As for the second problem, I've filed bug 909698 to track it.