1614388 - Enable automatic updating in China

Status: RESOLVED DUPLICATE of bug 1444406

(Core :: Networking: HTTP, task, P2)

Description

[u465659]

Attachment: 3.PNG

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:73.0) Gecko/20100101 Firefox/73.0

Steps to reproduce:

Open about window. And waiting about 5 minutes

Actual results:

Update checked and it updated failed

Expected results:

It should be updated to 73.0.b11

Comment 1

[u465659]

Attachment: 1.PNG

Comment 2

[u465659]

Attachment: 2.PNG

Comment 3

[u465659]

Attachment: 4.PNG

Comment 4

[u465659]

I'm So sorry about screenshot upload, This is my first time to use bugzilla

Comment 5

[Gingerbread Man]

If I'm not mistaken, "Update failed" means the update was downloaded but couldn't be installed. So this would be an updater or local configuration problem, rather than a server problem.

Comment 6

[Molly Howell (she/her) (mostly inactive)]

Actually it's the opposite, sorry. ;) That message specifically means that downloading the update file failed.

I'm afraid I don't have any solutions to offer, just one troubleshooting step: can you try pasting that URL that it's requesting (the one in that first screenshot, from Wireshark) into the address bar, and see if it times out again? If it succeeds, it should download a file with a .mar extension, but if it times out then that indicates some network problem, and I'll refer this bug over to the networking team.

Thank you for filing this bug, by the way.

Comment 7

[u465659]

(In reply to Molly Howell (she/her) [:mhowell] from comment #6)

Actually it's the opposite, sorry. ;) That message specifically means that downloading the update file failed.

I'm afraid I don't have any solutions to offer, just one troubleshooting step: can you try pasting that URL that it's requesting (the one in that first screenshot, from Wireshark) into the address bar, and see if it times out again? If it succeeds, it should download a file with a .mar extension, but if it times out then that indicates some network problem, and I'll refer this bug over to the networking team.

Thank you for filing this bug, by the way.

Thanks for replay

I tried the URL from Wireshark,When I use the SSL protocol(https://download.mozilla.org/******) to access this URL, it can download a file with a mar extension,But when I use the HTTP protocol(http://***),It prompts me the connection was reset.

So I guess that Firefox's update program requests 80 ports by default when it requests updates. If it uses 443port by default, everything will be OK

I try to use the extended "HTTPS everywhere"Make it mandatory to use HTTPS protocol,But I failed.

And I try requesting three IPs from nslookup

35.163.22.219

34.217.182.20

52.26.53.89

New problems found

When I try to visit 35.163.22.219 In the Chinese version(zh-CN) of Firefox, the page will jump to [https://www.firefox.com.cn/?utm_medium=referral&utm_source=mozilla.org]

When I try to visit 35.163.22.219 In the english version(en-US) of Firefox,the page will jump to[https://www.mozilla.org/en-US/]

I hope the above can help you,And I'm sorry that my English grammar is poor

Comment 8

[Molly Howell (she/her) (mostly inactive)]

(In reply to Wen Jie Ding from comment #7)

I tried the URL from Wireshark,When I use the SSL protocol(https://download.mozilla.org/******) to access this URL, it can download a file with a mar extension,But when I use the HTTP protocol(http://***),It prompts me the connection was reset.

So I guess that Firefox's update program requests 80 ports by default when it requests updates. If it uses 443port by default, everything will be OK

Yes, update files are downloaded over HTTP; we have bug 1444399 to investigate using HTTPS instead, but it's never been a priority. That definitely points to some network situation that affects clear text connections but not encrypted ones.

Given that information, I'm going to refer this bug to the networking team who can hopefully help investigate further.

I try to use the extended "HTTPS everywhere"Make it mandatory to use HTTPS protocol,But I failed.

Hmm, yeah, I don't think an extension would have any control over that request, since it's from Firefox itself and not a web page.

New problems found

When I try to visit 35.163.22.219 In the Chinese version(zh-CN) of Firefox, the page will jump to [https://www.firefox.com.cn/?utm_medium=referral&utm_source=mozilla.org]

When I try to visit 35.163.22.219 In the english version(en-US) of Firefox,the page will jump to[https://www.mozilla.org/en-US/]

I think this is the correct behavior; that IP address always redirects to mozilla.org, but the Chinese homepage has its own domain, so mozilla.org/zh-CN redirects to there instead.

I hope the above can help you,And I'm sorry that my English grammar is poor

Thanks for the info! And your English is fine, don't worry about it.

Comment 9

[u465659]

(In reply to Molly Howell (she/her) [:mhowell] from comment #8)

(In reply to Wen Jie Ding from comment #7)

I tried the URL from Wireshark,When I use the SSL protocol(https://download.mozilla.org/******) to access this URL, it can download a file with a mar extension,But when I use the HTTP protocol(http://***),It prompts me the connection was reset.

So I guess that Firefox's update program requests 80 ports by default when it requests updates. If it uses 443port by default, everything will be OK

Yes, update files are downloaded over HTTP; we have bug 1444399 to investigate using HTTPS instead, but it's never been a priority. That definitely points to some network situation that affects clear text connections but not encrypted ones.

Given that information, I'm going to refer this bug to the networking team who can hopefully help investigate further.

I try to use the extended "HTTPS everywhere"Make it mandatory to use HTTPS protocol,But I failed.

Hmm, yeah, I don't think an extension would have any control over that request, since it's from Firefox itself and not a web page.

New problems found

When I try to visit 35.163.22.219 In the Chinese version(zh-CN) of Firefox, the page will jump to [https://www.firefox.com.cn/?utm_medium=referral&utm_source=mozilla.org]

When I try to visit 35.163.22.219 In the english version(en-US) of Firefox,the page will jump to[https://www.mozilla.org/en-US/]

I think this is the correct behavior; that IP address always redirects to mozilla.org, but the Chinese homepage has its own domain, so mozilla.org/zh-CN redirects to there instead.

I hope the above can help you,And I'm sorry that my English grammar is poor

Thanks for the info! And your English is fine, don't worry about it.

Thank you very much for your help :-)

Comment 10

[Andy Grover [:grover]]

This sounds like a network problem, not a code issue.

Jim, do you anticipate progress on bug 1444399 any time soon?

Comment 11

[u465659]

Thank you for your attention. As far as I know, Chinese ISP cannot normally resolve any domain name of mozilla.org on port 80 (Include addons.mozilla.org/archive.mozilla.org..etc..).Because it is caused by a firewall (you can google "GFW")，When port 443 is used, it can be parsed normally.

Comment 12

[u465659]

(In reply to Andy Grover [:grover] from comment #10)

This sounds like a network problem, not a code issue.

Jim, do you anticipate progress on bug 1444399 any time soon?

Thank you for your attention. As far as I know, Chinese ISP cannot normally resolve any domain name of mozilla.org on port 80 (Include addons.mozilla.org/archive.mozilla.org..etc..).Because it is caused by a firewall (you can google "GFW")，When port 443 is used, it can be parsed normally.

Comment 13

[Nick Alexander :nalexander [he/him]]

Just tweaking wording: I wasn't clear if "unable" was "enable" or "un-able", meaning "disable".

Comment 14

[Rachel Tublitz [:rachel]]

(In reply to Andy Grover [:grover] from comment #10)

This sounds like a network problem, not a code issue.

Jim, do you anticipate progress on bug 1444399 any time soon?

Hi Andy,

I recently took over managing this team. We're going to talk about it as a team on 2/25, and I'll update back here with details once we do. If you need an answer before then, feel free to reach out to me directly, or just comment here.

Comment 15

[Rachel Tublitz [:rachel]]

(In reply to Rachel Tublitz [:rachel] from comment #14)

(In reply to Andy Grover [:grover] from comment #10)

This sounds like a network problem, not a code issue.

Jim, do you anticipate progress on bug 1444399 any time soon?

At the moment, we're targeting Q2 for 1444399. It's been prioritized, but not currently at the top of the list.

Comment 16

[BugBot [:suhaib / :marco/ :calixte]]

The priority flag is not set for this bug.
:grover, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 17

[Andy Grover [:grover]]

Ok cool. Let me assign to you Rachel, and you'll know who to reassign to?

Comment 18

[Rachel Tublitz [:rachel]]

Sounds good, thanks Andy. And just as a heads up, the https work looks like it's a bit more complex than expected. We'd still like to tackle it however we may have other projects that take priority for Q2.

Comment 19

[Rachel Tublitz [:rachel]]

:hectorz, for completeness here, would you mind providing an update around what you're seeing from the Beijing office related to updates? Ie, able to update or not/etc? We talked about this outside of Bugzilla, but just so we have all of our information together in one place around this issue for folks who are following the bug.

Comment 20

[Hector Zhao [:hectorz]]

(In reply to Rachel Tublitz [:rachel] from comment #19)

:hectorz, for completeness here, would you mind providing an update around what you're seeing from the Beijing office related to updates? Ie, able to update or not/etc? We talked about this outside of Bugzilla, but just so we have all of our information together in one place around this issue for folks who are following the bug.

Hi :rachel,

We're not seeing any unusual reports of users stuck at old Fx versions.

Anecdotally, I've seen similar errors occasionally but later attempts would succeed. If necessary, we can help with trying to reproduce this in home/office environment with debug prefs flipped on.

I'm also wondering if upload pings might provide some indirect signals.

Comment 21

[Rachel Tublitz [:rachel]]

I'm going to close this as a dupe of our larger effort to serve updates via https which is ongoing.