Closed Bug 1641923 Opened 5 years ago Closed 3 years ago

Google Meet video call hangs

Categories

(Core :: WebRTC, defect, P1)

78 Branch
x86_64
Windows 10
defect

Tracking

()

RESOLVED INCOMPLETE
Tracking Status
firefox78 --- affected

People

(Reporter: alex_mayorga, Unassigned)

References

(Blocks 2 open bugs, )

Details

(Keywords: nightly-community, Whiteboard: [wfh])

Basic information

Steps to Reproduce:

Join a Google Meet video call

Expected Results:

A working video call

Actual Results:

The Google Meet tab got hung up completely, had to start an alternate browser and join from there.


More information

Screenshot: (if relevant, please attach a screenshot or screencast to this bug report)

Profile URL: https://bit.ly/2zG5U5a

Basic systems configuration:

OS version: Windows 10 2004

GPU model: Intel(R) HD Graphics 620

Number of cores: 4

Amount of memory (RAM): 24GB

Thanks so much for your help.

Michael, does this look anything like what you reported seeing last week? I noticed in the perftrace (expand all) that it is shutting down a conduit.

Flags: needinfo?(mfroman)

It looks something like it in that it is deep-ish into webrtc.org code. The destructor that is stalled is doing:

IncomingVideoStream::~IncomingVideoStream() {
  RTC_DCHECK(main_thread_checker_.CalledOnValidThread());
}

It does seem like we're seeing stalls in webrtc threads at various places. I don't see anything stats related here like in the previous bug.

Flags: needinfo?(mfroman)

Some people are experiencing a ~10 second delay at the beginning of a call before audio flows, I wonder if this could be related. Edit: After regression testing by community member that could reproduce that delay, it looks like this is not related.

CalledOnValidThread() is called in many places. Thinking about it a little more, removing the stats poling, while the right thing to do for WebRTC stats, might exacerbate this issue. Paul mentioned adding performance markers to the WebRTC, this might not be a bad incentive to start.

Severity: -- → S3
Priority: -- → P2

I have the same issue where Google meet just hangs (sometimes not only that but firefox crashes).
Using firefox-nightly 79.0a1
Basic systems configuration:
ArchLinux
CPU: Intel(R) Core(TM) i7-8565U
GPU model: Intel(R) UHD Graphics 620 (Whiskey Lake)

This happens only when using my laptop and it doesn't happen if I have it plugged or set the intel_pstate to performance via:
sudo x86_energy_perf_policy -d --all balance-performance. In windows this could be done by setting the energy saving option to performance (not 100% sure here, just a guess)

I will see if I can get a log too to ensure it is the same issue.

¡Hola Diego!

Is this still crashing for you?

If so, please provide us with a few of the crash signatures in the about:crashes page of your Firefox.

¡Gracias!
Alex

Flags: needinfo?(guida.forneas.diego)
Blocks: webrtc-perf

Setting this to P1 for investigation and analysis so we can determine its true impact to the user.

Priority: P2 → P1

I'm marking this resolved/incomplete since it has been quite a while since we've seen any activity here. Please reopen if it is still occuring.

Status: NEW → RESOLVED
Closed: 3 years ago
Flags: needinfo?(guida.forneas.diego)
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.