Add a meeting Rate this page

A

Okay, okay, so ready to start hello to this December edition I. Think the last one of 2017, that's been a pretty good year. Welcome everyone. Thank you for joining we're going to go through a quick update of a few things. We've been doing in the past couple weeks and then we're going to do open, mic and pulp. Whatever anyone wants to talk about so George, why don't you start yeah? Why don't I start so two things mostly the one is we fix.

A

The screen sharing issue reported in the community in the last community call by Jonas and Marcus. So that's now, fixed and I. Think it's not know I think I'm pretty sure it's also deployed in New Jersey. So we should not have that anymore. They've also reported well not reported, but do you want to take a couple words about what was going.

B

On there.

A

Very quickly, that was something that we had observed in the past, but we were not able to identify the good cause we have. We have identified symptoms. Actually, the chrome was failing to decrypt some artistic messages coming from the bridge that that has been we have. We have found that a long time ago we have looked into that with Boris, but it wasn't really forensic the cause so and the conference was working so yeah.

A

We got swamped with other things and we didn't investigate it further, but it turns out that actually, the bridge was not properly terminating a fires and please- and so it was it it may it may be they.

A

It may have been the case that forwarded please and in FA are some reason where were were failing to be decrypted and they were not being generated by the bridge because they were not being detected incoming please, so they were not being terminated, they were just being forwarded, but in a in a malformed state, so the client often failed to decrypt the incoming fi RM. Please so yeah well was pretty much that, but could have resulted in unexplained connectivity issues and freezes during during the call.

A

So we are very happy and thanks very much much Jonas for unn Marcus for bringing this to our attention, because I believe this has been gratifying- calls for a long time in our in our system in our platform, but now that's fixed and we should be. We should be good what else I'm also I'm, currently working on adding some packet tracing login packet, traces from our infrastructure. This is in order to be able to to enable us to improve the GCC algorithm in some in some sense.

A

So maybe you'd like to point out explicitly that we're not talking about tracing content, only metadata.

C

Yeah.

A

Just delivery times and debugging state information from the GCC from the congestion control algorithm that we are we're using. This is not going to log any content, of course, and there's also another issue that Jonas and Marcus has been trying to figure out, and this is again really with respect to please and the rest interface so sometimes when it might occur that the bridge is not generating a plea for when the participant and the participant joins the call this is when using the rest interface again.

A

This has been known for a while, but we haven't been able to reproduce it in our system. Hopefully, with their help. Now we are going to be able to reproduce it and solve it. I'm gonna have a look at that during this week. Yeah, that's that's what most of what I had to say, thanks George Boris.

C

Just a note about analytics we're restructuring things, so all of the client analytics events or most of them are gonna change, their format and the adapter API that we have in the library changes its API a bit.

C

So what this means for people is, if you have your own analytics handler implementation, so something that pushes events to your own back-end you're gonna need to update it.

C

I'm not aware of anyone using something like this, but I'm, putting it out there for people. So we know we're. Gonna have documentation on the API, so you can. It should be an easy change if you're using Google Analytics we're gonna make the current we're gonna update the current implementation ourselves, but the format of the event is going to change so you're going to see changes in your dashboard and.

A

Very quickly, by the way, is anyone you're using YouTube, with no analytics.

C

Or with their own implementation,.

A

Feel free to post in the chat, if you, if you'd like later, ok thanks Boris, oh I, see you didn't raise your hand, so I guess I'm going to quickly. Do you want to talk a little bit about the the ipv6 issues that we've been noticing.

D

Yes, so it turned out that our Android app will not work on a 56 Network.

D

One reason is that on the bridge we don't have ipv6 and the ice will not work when we're passing only ipv4 candidates by adding ipv6 on our roadmap right now. So.

A

You because we're talking about me jizzy the mediated appointment, not didn't meet itself. Yes,.

D

Which.

A

Already has no problems with IP visit. Yes,.

D

And another.

A

Thing.

D

Is that if you're, adding or turn instance I reversed and just make sure to use the domain name instead of like ipv4 address directly, so that it will also work on ipv6, because then it will resolve this domain name to ipv6 name address. Otherwise, I found that ipv4 addresses were not working.

A

Right so I'd like to expand on this a little bit, because it is something that I find strange. So what we found out in the past is that if you are on an ipv6 only network that has not six four capacity, you things do end up working well in our, so they do actually I'm not even sure what this situation would be on web, probably things we just fail there, but so on mobile app.

A

Oh, do this testing as part of their admission process, where they check whether the out actually works in a in a six, only Network we're not six four capability, and they actually, we had a rejection very early on in analysis that then Luber resolved by basically making sure that whenever we deal with an ipv4 address, we're going to try and convert it into a valid ipv6 address by using some native value. S call that, based on that six, four does that translation?

A

So that's been working well, but what pub was referring to is that unless you have a DNS name- and you led the system- the conversion through as part of the DNS routine, which would work fine unless you do that, we're not able to find on Android a way to convert an ipv4 address into six one using not 64, so I'm I'm wondering if anyone else here has worked through that issue and found a way to be to do these conversions when Android.

A

Let me summarize the question again: has anyone found a way on Android to convert an ipv4 address to us? I, give you six one uses in not six one.

A

If someone decides, they would like to share that later on, please don't hesitate to jump in because that's of interest but yeah until we'd still have to still have to go in and have to talk her that were station are. You is that we might just say that this would only work if your infrastructure only also has administered okay thanks pal, so you said, I should consider your hand raised.

E

Okay hear me: we.

A

Do.

E

Well, that much to report other than I was working on error handling for the most part of these two weeks. So I guess this will be available. The next, with the next update of the app on the store and.

A

To be clear, when you mean that you're handling with what you were you're talking about, it is properly displaying disruption is in connectivity to the user.

E

Yeah, it means a number of things, so it means that if they we failed to join the MOOC. If the conference fails, if the connection fails, it will get like an overlay like on the web. So a page, this is hey, there's a problem, we're going to retry now we're a bit or you can cancel or trigger it by hand with a reload. That's one side, I know there's another part of this F word, which hopefully lands that today or tomorrow, is rendering information together with the Avatar.

E

When the user has a problem, for example, his client crashed or something and we stopped receiving media for him or the bandwidth is too low when he was suspended and right now, I'm working on detecting the case when remote doesn't send any video, but ice is up and everything looks fancy, but video is not being received, but usually this explained by connectivity issues or something I like, but that requires a patch to the react native web RTC library, because this is this.

E

Detection is not built into web RTC, which is implemented in Chrome, so I'm kind of taking some work that patented and they in turn, are in the iOS part.

A

Right great, thank you very much is all you on the call.

A

Okay, I'm, going to assume that he's not here so is all you recently started. Work on history feature absurd to recall the recent calls list. So now, when you start the app you're going to see a bunch of conferences that you joined in the past and be able to quickly rejoin them, which would be particularly helpful, especially if they were another servers.

A

So this way you won't have to type the entire name again and on that matter, where he's next going to move to a settings window which is going to let you specify your own server so that, if you're using a server rather than the agency you can, you can have a better experience with.

A

With the mobile app now, I lost my train of thought there.

A

So, basically, the way that this would work is that, if you specify your own server would still be able to join other servers, but you would have to type their full URL, including from egt and for your own server and you're, just going to type the conference name, the behavior that you get today for the agency. So that's something that people have been confused by in the past and that setting stage is also going to let you set your name in Avatar.

A

So.

A

So Edie has a question about at five yards and feel eyes, but beef George. Would you like to take that or maybe ad would you like to or maybe I would you like to say exclusively WOD what you're interested I.

F

Am I joined the call a little bit late, so I didn't get the context, but I heard something about FR, f ir p li and how it's been a problem for some time, been crap defying or find a service. So I just wanted to learn a bit more in.

A

That lori talked.

F

About.

A

Yeah that that was, let me give you a quick version and George you jump in if you, if you haven't been accurate, but basically the problem is that plin fi, our termination wasn't working properly, so fi airs in PL is sent by one client, we're actually not terminated and leaked through to other clients and the encryption, and they weren't probably properly being forwarded. So they were failing as what was that cruel message. It was uh unprotected unprotect failed. So that is is what we recently fixed. George.

F

Very interesting we've been noticing those messages for quite some time, so that was.

A

The biggest they, basically, it wasn't an encryption problem per se, but just a forwarding problem. So now these are not being forwarded anymore and we're probably properly terminating them. So we wish you our own if ir doesn't realize there George. Would you like to correct anything? I said that law was pretty pretty accurate description.

A

Ed. Does that answer your question yeah.

F

Yeah and I'll be looking to bring in ethics.

A

Cool okay: what would you like to add something on the mobile side.

G

No, that's my show and working some fixes in there.

A

Okay,.

A

So I see another question from Roy. The first one was about a comparison between zoom in GT. We haven't done that. We've done one in the past over blue jeans. I, don't think we're planning on doing something like that in the short term. But if someone wants to do it would be happy to to talk about ways on how this could be achieved.

A

The second question from Roy is about messages that now show on mobile, saying that video has been disabled to save bandwidth.

A

So Rory should I understand that you feel that these messages are improperly shown when there should be in the bandwidth there.

H

Should be enough bandwidth, and actually this was on desktop Oh.

A

Was it.

A

So we so we're using this on stable on your own deployment.

H

Stable our deployment there, a a WebEx and zoom and Atlassian customer we're trying to move them over to you, but several of their users will occasionally see that screen where we didn't see that before on the previous stable release, that's something that we did or is there a change on your side, George.

A

Has the same side manifestation made it in this table? Do you no no.

C

No, no.

A

It's not, and even if it did make it even if the code did make it into stable, it's not enabled by default, so so yeah that that wouldn't explain the issue. But we're trying to remember whether we have any fixes.

B

Yes,.

H

Yes, yes, I.

B

Was wondering it may be 1080p Chrome good effective because technically we have to depend ATP, the lowest resolution is going to be higher than it used to be, and they don't do. Resolution drops in simulcast. So maybe that's something.

A

Are you talking about that old bug, Brian, no.

B

I'm just saying that when we change the chrome to do 1080p now with simulcast, the lowest layer resolution is higher than it was before, and maybe that's is imposing a slightly higher bandwidth concern. You know requirement or something because they don't do. Resolution drop something simulcast layers, I, don't know if they'll bring the bitrate all the way down hill or not bill. Maybe that's.

A

Interesting.

A

It.

H

Does we hadn't seen that message before so.

A

What what Brian is saying is that in the latest version, one thing that changed is that now the maximum resolution is 1080p as opposed to 720p, and when that happens, even the lowest resolution that we can potentially drop down to in order to adjust the traffic problems also goes up, so our minimum goes up, which means that it is more likely that, rather than dropping into a shittier resolution, we would just cut off here cut off your video.

B

Okay, this.

H

Is Brian yes, great I'll. Send you a note, Brian I we'd like to know how to set that. Is there a custom to WebEx which has well? If you look at the comparison that zoom did with WebEx the WebEx resolutions, not very good, so they're, not that kind concerned about it, but if they drop completely off, obviously that's a hindrance.

A

I'm, sorry, can you repeat that.

H

Jitsi resolution is very good in comparison to the certainly to WebEx, but if they drop completely off and they get that message that they're been with their video has been turned off to conserve bandwidth. Obviously, that's a negative in their mind, sure.

A

So I sent you in the chat here a link to the comping JSON me jizzy. You can see how to configure it to two 720p and you can experiment with that a little bit okay.

A

So this should drop all resolution layers, including the lowest one, so we should be sending shittier quality video for longer before abandoning does that make sense. Yes, thank you sure. Let us know how this works.

A

So, let's see speaking of Bryan we're still making progress on implementing support for for error correction in the bridge. We have we're now almost done with the receive port I'm, not sure.

B

Okay,.

A

See that's pretty cool, if do you know if it's unstable, I.

B

Don't.

A

And by the way, it's not enabled by the phone in CRO anyway, so right.

A

Bryan is also currently working on sin side forward. Error correction. Is it would you like to add anything to that Bryan yeah.

B

Just early I'm starting I'm, just taking a look at what that chromes implementation as a reference point and then I'm gonna start doing that. But, as Emile said, this is still behind a field trial in Chrome, but we're hopeful it will make it through at some point soon, because the current documentation is not not really useful. That's why nobody's using it. So with this we would be able to get a real one that would offer some nice protection and maybe be able to make probing and things a lot more interesting.

B

So there's a lot can be a lot of upside to this one that gets shipped everywhere.

A

Okay, I think that does complete a a good round here. um Anyone on the team wants to add anything in terms of professed.

I

I guess we've launched a couple new regions. That's got some news all.

A

Right we.

I

Now.

A

Have.

I

Shards of meat at sea in Frankfort, as well as Singapore, so we're going to view are coming in from Asia or Central. Europe you'll have the potential to be hidden, big new shards. So far, we've seen quite a bit of bandwidth, hitting both so I'm, pretty pretty hopeful that this is improved. Our service overall.

A

Yep thanks iron.

A

So, unless I miss is something I think that completes the progress part on our side and we're now moving into open mind any questions, comments or requests or discussions.

J

Hi guys either makers I've got a quick question, I'm not running on the bridge code, and you know of me and sorry: the meet code. That's a couple of weeks, maybe a couple of months old but I'm, hoping that this is an old book that you already fixed but we're running into an issue where, when people mute themselves for a longer time and then try to unmute them, that visibly works, but the audio never gets restored. Is that something a problem that you read? Sorry.

A

Can hear it you were cutting up at least for me. I'm sorry, can you repeat the question we're.

J

Running into an issue where people that are muted and then unmute themselves, you never hear them. Did you see them? The mute was only when they unmute themselves. The audio doesn't come back. How.

A

Old is that version a.

J

Couple of months, probably.

A

Deserves an eclipse, same problem, kill anyone.

A

It can be a lip sync issue because we have it disabled by default. No, but what if they had enabled in.

J

What's that telling is a yes.

A

Yeah yeah, mostly, you can optionally disable that in the bridge as well, but it does have a complicated option. All.

J

Right enable live, see, ya, I'll, have look there now, try to update and should.

A

Be disabled.

J

Okay, I'm.

A

Joining from Chrome yep or any ideas as to what else might be causing this I would look for.

A

No I was about to say to look for it remote description errors, but for audio we we don't do sit remote descriptions.

B

Grab for the audio stream show the activity or any indication that it started at that point, I.

J

Haven't I was just hoping that you guys immediately recognized this problem and said: oh just a bit and it'll be gone. I'll update, anyways and I. Look at the lip sync settings and if it persists so I'll bring you guys back in the making a mess or something I would.

C

Also, take the chrome logs for failed to and protect packet errors in case that the gap in sequence numbers. Oh.

J

Thanks.

A

Oh, that's a that's a good one. Yeah also.

C

Shouldn't be happy.

A

We've.

C

Seen problems like a few times, but we've I as far as I know, we've fixed them all.

J

Okay: okay, thanks! Oh one other thing: I tried to join using the Android that I just downloaded and ia allowed me to connect and briefly join the conversation. But then it kicked me out repeatedly at three or four times, and that happened I think. As soon as the presenter changed there was some flickering and image and then I application, yeah.

G

If you're quick enough to going to our your mouth, it shouldn't cars, Sonam crash, we fixed it in master thanks for reporting, was.

A

It crashing dog, uh sir, were you with the exact experience of the.

J

Application crash completely: okay,.

A

Right so this should be coming out in the next update, which you would you like to venture a guess as to when we'll do that.

J

Okay,.

A

Okay, you had a question.

K

Yeah actually I wanted to talk about gaseous problem with audio. Do you guys hear me? Well, yes,.

C

Okay,.

K

I also noticed the same issue: Gus did your. Did you notice the issue that is in a longer session somewhere around two hours or something like that? I thought longer.

J

And the person is that it happens today tend to be muted for a longer time, but that's not two hours. Our meetings are about an hour tops and those people would be muted. For, like ten minutes, okay I, we haven't really experimented it with it to find the boundary values, but it it appears to occur when people are muted for an extended period period.

K

Yeah, we have not just notice the same similar issue when session is long longer like as I said two hours or something like that, so apparently jbjb sending packets, but they never actually get processed by the bar TC stack, so I'm not sure, what's happening there, but the something something to investigate for sure. Well,.

A

That actually sounds like, like.

A

What's the word, a sequence number, what's the word for the rollover.

K

Right.

A

Yeah.

K

It does look like that, so.

K

I'll probably take a look at it, I'm not sure, but I.

A

Think we've actually addressed that one. What what version have you seen that way that magic.

K

It was quite somewhat somewhat too old.

A

Can someone confirm that we did look into that and address it at some point? So we have a code that hides the gap to the audio sequence, numbers I I think we have. It may be the case that we have disabled it. So I can check when you put where there yeah George, because we because this messes up with lip sync and at some point we're trying to fix lip sync and one of the efforts was to one of the ideas- was to disable this hiding of the gaps. Let me check.

A

I will come back with an answer.

C

Somewhat easily confirm whether this is the problem by so the threshold is somewhere just about 10 minutes between 10:00 and 11:00. So I would need for 12 minutes and then look at the logs from chrome, the debug logs. And if that's the problem, you should see a bunch of error messages saying that the Pockett failed to unprotect. Okay,.

J

It might be worth noting that I'm running your guys, rep your guys code, wrapped in my own code, so my or.

A

Then itself.

J

No of it being now my fault, so don't take my word on that's just a side course. You don't want to do that.

A

No, but you also said around 10 minutes, so that's a suspicious screen, sealants yeah.

A

While George is looking for an answer to that thing, anyone else wants to open a subject.

J

A quick question is any of you going to attempt to force them in Brussels in February.

A

El segundo.

J

Hominem there.

A

Well said, rule is actually on Disko, so so would you liked it.

J

Now I used I think he's the lead of the real-time lunch. Even so, I expected him to be there. I.

C

Think Nick from our team, the Nick vests and who did the speech-to-text stuff might be there I think he's giving a talk.

D

Yeah and I'm not sure about so.

J

You know because if anyone's there come visit us we're all in the real time, communication is large. So.

A

When we hear when you hear you know.

E

Okay, sorry I'm with a debug build fighting for all the debug messages. I know: I'm, not gonna, be there this year, I'm on the backstage ah currently browsing through the talks. A lot.

A

Of responsibility for a nickname, yes,.

E

Own ink is gonna, carry the flag, we accept his talk, I'm, not sure yet you know and yeah. This talk is about work and speech recognition. Now we have some great talks this year by the way, um some new new interesting topics as well, a bunch of people that have not submitted togs of previous years, which is great to see some refreshing in the room. So if you, if you're there definitely go check it out.

A

Okay, George, you ready with me answer: it's.

A

Okay, so why don't we take this to the meaning list and and because, if it's not, if that's broken, we'd have to go and have to create an issue? Anyways yeah we're also going to an open a ticket.

C

Okay last go.

A

A.

C

Label.

A

I.

C

Don't think we're alive. Oh.

I

Yeah we were live-streaming but I didn't see it fail. I'll take a little bit log see what happened.

A

Thanks also, it looks like it's enabled by default, so either that's broken or something else is going on. Oh we, it will be great for whoever reported these muting issues really great for you guys to report that you're having them on a recent version and- and then, if you are we're, we absolutely need to take a look into this allah-allah.

J

Version and then I'll, let you guys know: okay.

A

Thank you very much everyone for joining and see you again next year.

A

You.

A

You.

A

Oh you.

A

Oh.

A

You.

A

You.
youtube image
From YouTube: Jitsi Live Stream

Description

No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).