648967 - If 'Check for Updates' is clicked and fails to reach server, it will say Firefox is up to date

Status: RESOLVED DUPLICATE of bug 679742

(Firefox :: General, defect)

Description

[Tony]

User-Agent:       Mozilla/5.0 (Windows NT 5.1; rv:2.0) Gecko/20100101 Firefox/4.0
Build Identifier: Firefox 4.0

Firefox -> Help -> About Firefox -> Check for Updates
Checking for updates...
Firefox is up to date

Firefox has failed to check whether it is up to date while saying that it has.

Reproducible: Sometimes

Steps to Reproduce:
It happens when it fails to reach the server.


Expected Results:  
It should say that it has failed to reach the Update server, not say that it is up to date.

Comment 1

[(mostly gone) XtC4UaLL [:xtc4uall]]

(In reply to comment #0)
> Steps to Reproduce:
> It happens when it fails to reach the server.

How do you know it *had to* fail so that above Assumption would be correct?

Comment 2

[Tony]

I am watching the attempted HTTP connection on <a href="https://addons.mozilla.org/en-US/firefox/addon/httpfox/">HttpFox</a>, and I see that the request failed yet then the "About Mozilla Firefox" box displays <i>Firefox is up to date</i>

Comment 3

[(mostly gone) XtC4UaLL [:xtc4uall]]

Hmm, ok.

Please set "app.update.log" to "true" in "about:config" and check the Error Console Output after the next Update Check what has to fail and post it here.

Comment 4

[Tony]

(In reply to comment #3)

AUS:SVC Checker:getUpdateURL - update URL: https://aus3.mozilla.org/update/3/Firefox/4.0/20110318052756/WINNT_x86-msvc/en-US/release/Windows_NT%205.1/default/default/update.xml?force=1

AUS:SVC Checker:checkForUpdates - sending request to: https://aus3.mozilla.org/update/3/Firefox/4.0/20110318052756/WINNT_x86-msvc/en-US/release/Windows_NT%205.1/default/default/update.xml?force=1

AUS:SVC Checker:onError - request.status: 2152398861

AUS:SVC getStatusTextFromCode - transfer error: Connection refused, code: 2152398861

UTM:SVC TimerManager:notify - notified @mozilla.org/browser/search-service;1

Comment 5

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

The URL in comment #4 returns the correct empty update for me. Smells like a network/firewall issue on the machine running Firefox. 

That said, it seems strange that the about box says 'Firefox is up to date'.

Comment 6

[Robert Strong (they/them - no direct email)]

I *might* be misremembering this but iirc beltzner wanted to go with the 'Firefox is up to date" when encountering errors when checking for updates in the about box.

Comment 7

[Tony]

(In reply to comment #6)
> I *might* be misremembering this but iirc beltzner wanted to go with the
> 'Firefox is up to date" when encountering errors when checking for updates in
> the about box.

I don't understand why this should be the case.
In UNIX, if everything is ok, you get no message.
With this feature, when an error occurs, you get a message saying everything is ok.

Comment 8

[Robert Strong (they/them - no direct email)]

There are several cases where an error can occur with checking for an update and everything is actually fine due to the nature of the internet. Also, there is code that checks if the problem is re-occurring without ever being successful in which case the user is notified.

Comment 9

[Robert Strong (they/them - no direct email)]

This decision by the UX team not to show an error message was intentional. Also, this UI is actually part of the Firefox product and changing the behavior would not be done in app update. Moving over to Firefox -> General

Comment 10

[Tony]

(In reply to comment #8)
> There are several cases where an error can occur with checking for an update
> and everything is actually fine due to the nature of the internet.

error means everything is fine?
nature of internet?

> Also, there is code that checks if the problem is re-occurring without ever 
> being successful in which case the user is notified.

I have never seen this, how do force this notification so I can see it?

Also, how many times does the error have to occur before the user sees that
there is a problem. The reason that the user would manually check if the
browser is up-to-date is because they want to make sure that they are running
the most up-to-date version so that their browser won't be exploited using a
publicly available security vulnerability.

(In reply to comment #9)
> This decision by the UX team not to show an error message was intentional.

I would really like to see the rationale behind this decision, was it discussed in a forum or blog post?

> Also, this UI is actually part of the Firefox product and changing the
> behavior would not be done in app update.

Not a problem, you can just include the fix in either Firefox 5, 6 or 7 this year.

Comment 11

[Robert Strong (they/them - no direct email)]

(In reply to comment #10)
> (In reply to comment #8)
> > There are several cases where an error can occur with checking for an update
> > and everything is actually fine due to the nature of the internet.
> 
> error means everything is fine?
> nature of internet?
Yes

> > Also, there is code that checks if the problem is re-occurring without ever 
> > being successful in which case the user is notified.
> 
> I have never seen this, how do force this notification so I can see it?
It isn't easily tested manually and we do have tests for this. The tests are available at:
http://mxr.mozilla.org/mozilla-central/source/toolkit/mozapps/update/test/chrome/

I believe test_0121* through test_0151* are applicable

> Also, how many times does the error have to occur before the user sees that
> there is a problem. The reason that the user would manually check if the
> browser is up-to-date is because they want to make sure that they are running
> the most up-to-date version so that their browser won't be exploited using a
> publicly available security vulnerability.
When there are 5 background checks that fail in a row. It is reset if there is a successful background check.

> (In reply to comment #9)
> > This decision by the UX team not to show an error message was intentional.
> 
> I would really like to see the rationale behind this decision, was it
> discussed in a forum or blog post?
I believe the bug is the only place it was discussed - bug 596813

> 
> > Also, this UI is actually part of the Firefox product and changing the
> > behavior would not be done in app update.
> 
> Not a problem, you can just include the fix in either Firefox 5, 6 or 7 this
> year.
It is up to the Firefox UX team or Firefox product drivers whether the behavior should change.

Comment 13

[a_geek]

I have started to see this problem on Linux after FF 8.0, when it consistently failed to find the updates to 8.0.1 and 9.0. That can't possibly be an effect of network states or plugins, because I eg. today just immediately before update FF 3.6 to 3.6.25 and being greeted by a message that I should download 9.0 instead (= no network problems), and when checking for updates in FF 8.0.1, as root and after rm -fr ~/.mozilla (= no plugins), it still came up as saying that my 8.0.1 is up to date. I tried several times, then went for manually downloading and installing the 9.0 version. I'd prefer to see this bug re-opened, and/or app.update.log = 1 by default. Updates don't happen that frequently as to clutter my disk, do they?

Comment 14

[dokc]

I had the same problem on windows using FF 4,5,6,7,8,9. Finally I found the cause of the problem. Firefox tries to update using URL from app.update.url
In my case (https://aus3.mozilla.org/update/3/%PRODUCT%/%VERSION%/%BUILD_ID%/%BUILD_TARGET%/%LOCALE%/%CHANNEL%/%OS_VERSION%/%DISTRIBUTION%/%DISTRIBUTION_VERSION%/update.xml)

The problem is that my security settings requires that I always confirm untrusted https connection. Unfortunately https://aus3.mozilla.org was untrusted. I just needed to paste the address once in FF, add exception and check updates again. The new update was available in about box.

Comment 16

[Thomas Ahlblom]

Reproduced (by putting "127.0.0.1 aus3.mozilla.org" in /etc/hosts, and also with an IP on the internet which is blocked by a firewall):
Mozilla/5.0 (X11; Linux x86_64; rv:13.0) Gecko/20120222 Firefox/13.0a1

Comment 17

[Thomas Ahlblom]

Can someone please move the current bug to a suitable component so we get an authoritative decision whether Firefox should lie about its update status or not?