1586423 - meet.google.com / Google Hangouts doesn't work in Nightly ("Couldn't start the video call because of an error")

Status: VERIFIED FIXED

(Core :: WebRTC, defect, P1)

Description

[Daniel Holbert [:dholbert]]

STR:

1. (optional) use a fresh Firefox profile
2. Start a meeting at https://meet.google.com/ , or visit https://meet.google.com/jkf-gpzy-owq (a meeting I started; not sure how long it'll be active)

EXPECTED RESULTS:
Underneath the permission prompts, you should see a black video preview area and "What's your name"

ACTUAL RESULTS:
The page instead says "Couldn't start the video call because of an error". (There are still permissions prompts for my video/audio, but I'm still unable to join the meeting even after I accept these prompts with the "remember" checkbox.)

This seems to be a regression from bug 1470568.

mozregression gives me a regresssion range of https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=a4a3a37a8a9499865c7dbc8a2678154317af9dc1&tochange=2fca72b71c56c2cf8bcc736f814aa8f76ad3382d
which just includes that one bug.

Comment 1

[Daniel Holbert [:dholbert]]

[sorry, I hit a Bugzilla bug and it lost my first-comment comments on bug-filing, grr. I'll retype them in comment 0 via "Edit" pencil-icon.]

Comment 2

[Daniel Holbert [:dholbert]]

[Tracking Requested - why for this release]: breaks a relatively-important video conferencing site

Comment 3

[Andreas Pehrson [:pehrsons]]

whereby.com is also broken when used with their SFU.

SfuConnectionManager: failed to create answer for offer DOMException: "Failed to parse SDP: SDP Parse Error on line 21:  Warning: Unrecognized attribute (rtcp-rsize) 
SDP Parse Error on line 40:  Warning: Unrecognized attribute (rtcp-rsize) 
SDP Parse Error on line 25: No rid attribute for 'rid=hi'
" 3ec61eca-da75-4497-90cf-a2d29deaa062

Comment 4

[Byron Campen [:bwc]]

These services are using a format that has been invalid for more than 3 years (since draft-ietf-mmusic-sdp-simulcast-04), so it is about time to switch over to the new format. Who are our contacts at meet and whereby nowadays? I could maybe see reverting this change temporarily to allow more time to fix this, but it is not a very complicated fix.

Comment 5

[Andreas Pehrson [:pehrsons]]

I've reached out to whereby, but it's unclear when they can have a fix for this.

Comment 6

[Nils Ohlmeier [:drno]]

I reached out to the Meet team. I'll update once I hear back from them.

Since we are pretty close to a train ride I think we should consider reverting the change from bug 1470568 for 71 and consider doing it at the beginning of the 72 cycle.

Comment 7

[Byron Campen [:bwc]]

Ok. Let's see where we are next week on fixes from meet and whereby, in case they can fix their end in time.

Comment 8

[Henrik Mitsch [:hmitsch]]

I suggest extending the Bug Title
from
meet.google.com doesn't work in Nightly ("Couldn't start the video call because of an error")
to
meet.google.com / Google Hangouts doesn't work in Nightly ("Couldn't start the video call because of an error")

It took me a couple of days to find this bug because I was not smart enough to use the right keyword. Hope this minor change is okay with people here.

Also, thanks to the people who are working on fixing this. You are stars!

-Henrik

Comment 9

[Henrik Mitsch [:hmitsch]]

PS: Here's my workaround: I installed a release version of Firefox and run this in parallel to Nightly to keep my Google Hangouts going until this bug is fixed.

Comment 10

[Byron Campen [:bwc]]

Have we heard back from whereby and meet about this?

Comment 11

[Kelsey Gilbert [:jgilbert]]

This sucks a lot from a user perspective. Can we delay breakage until after we have traction on ensuring major application are updated?

Comment 12

[Andreas Pehrson [:pehrsons]]

My contact at Whereby is on leave until end of October, and no-one else there has picked this up AFAIK.

Comment 13

[Kelsey Gilbert [:jgilbert]]

Attachment: Bug 1586423 - Revert bug 1470568 for breaking Google Meet, etc.

Comment 14

[Nils Ohlmeier [:drno]]

(In reply to Byron Campen [:bwc] from comment #10)

Have we heard back from whereby and meet about this?

The Hangouts team replied back today that they are encountering problems with Firefox when trying to handle it on their end. So I think we should revert this to prevent this riding into Beta and give services more time to adapt.

Comment 15

[Nils Ohlmeier [:drno]]

(In reply to Jeff Gilbert [:jgilbert] from comment #11)

This sucks a lot from a user perspective. Can we delay breakage until after we have traction on ensuring major application are updated?

In general that would be ideal. But it is completely unpredictable if and which service out there is going to have problems with our changes. Which would then mean we would have to test every single service out there before we land patches.

Comment 16

[Andreas Pehrson [:pehrsons]]

Did we issue deprecation warnings for sites trying to use rid= before removing the support? Doing that for a couple of versions prior (and including both deprecation and removal in the release notes for devs) seems like a best practice. At least then sites cannot come and say we didn't warn them.

We could hook a telemetry counter to the deprecation warning too, to have an idea of the damage we'll be causing.

Comment 17

[Kelsey Gilbert [:jgilbert]]

Yeah, we have processes for use-counting to probe. It's fine to break things on accident sometimes, but we should rush to revert once we get a clear regression like this.

I've been unable to use Google Meet (which I need for work calls with other companies!) in my default browser for over a week. It's been almost two weeks without resolution, let's revert this.

Comment 18

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/30a9a8619ce1
Revert bug 1470568 for breaking Google Meet, etc. r=bwc

Comment 19

[Stefan Hindli [:stefan_hindli]]

https://hg.mozilla.org/mozilla-central/rev/30a9a8619ce1

Comment 20

[Kohei Yoshino [:kohei]]

Updated the site compatibility note for Bug 1470568.

Comment 21

[alex_mayorga]

¡Hola!

https://meet.google.com/ works again on Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:71.0) Gecko/20100101 Firefox/71.0 ID:20191017093524

Should this one be linked to https://bugzilla.mozilla.org/show_bug.cgi?id=1431543 for visibility?

¡Gracias!
Alex

Comment 22

[Kelsey Gilbert [:jgilbert]]

Yes, thank you.

Comment 23

[Andreas Pehrson [:pehrsons]]

Whereby should be fixed now FWIW.

Comment 24

[Alex Burke]

Firefox 72.0 (at least on macOS) has broken Meet:

SDP Parse Error on line 36:  Warning: Unrecognized attribute (x-google-flag) 
SDP Parse Error on line 60:  Warning: Unrecognized attribute (x-google-flag) 
SDP Parse Error on line 30: No rid attribute for 'rid=f'
SDP Parse Error on line 30: No rid attribute for 'rid=f'```

Comment 25

[Byron Campen [:bwc]]

(In reply to Alexander Burke from comment #24)

Firefox 72.0 (at least on macOS) has broken Meet:

SDP Parse Error on line 36:  Warning: Unrecognized attribute (x-google-flag) 
SDP Parse Error on line 60:  Warning: Unrecognized attribute (x-google-flag) 
SDP Parse Error on line 30: No rid attribute for 'rid=f'
SDP Parse Error on line 30: No rid attribute for 'rid=f'```

Can you attach the SDP here?

Comment 26

[Alex Burke]

Of course — could you let me know how to capture it?

Comment 27

[Byron Campen [:bwc]]

You could run Firefox with "MOZ_LOG=signaling:5,jsep:5" as an environment variable, or you could install the addon described here:

https://blog.mozilla.org/webrtc/new-tool-debugging-webrtc/

Comment 28

[Alex Burke]

Attachment: webrtcDevtool.html

Nothing appears in this pane.

Comment 29

[Alex Burke]

Specifically, the behaviour I'm seeing is identical to comment zero; this worked fine in ff71, and also works fine today (right now) in ff68ESR.

Comment 30

[Alex Burke]

I have filed bug https://bugzilla.mozilla.org/show_bug.cgi?id=1607781 related to this.

Comment 31

[Saeed Jahed]

For reference, the RID= SDP issue was fixed in Meet/Hangouts in late October.

Comment 32

[Ryan VanderMeulen [:RyanVM]]

Looks like we're tracking the Fx72 issue in bug 1607781.