[jitsi-dev] Stability issues


#1

In my product, that wraps the videobridge, focus component and jitsi-meet,
we're running into strange stability issues.

A one-on-one conversation works fine - we've been talking for ages without
incident. A third participant does not appear to cause problems either, but
when the fourth participant joins, people are starting to have client
crashes all the time (like once a minute or something). The interesting bit
is that this did not affect the user that was using Chrome 55 (who was the
first participant). The others were all using 58.

I've seen connectivity issues during "the Call" as well, as have I heard of
various references to browser bugs. Are my issues related? If so, what
exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?

Regards,

  Guus


#2

Instant follow-up: A curious thing was that all stability issues
disappeared, when participants three and four left the conversation. After
that, things were back to normal.

  - Guus

···

On 1 June 2017 at 11:58, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:

In my product, that wraps the videobridge, focus component and jitsi-meet,
we're running into strange stability issues.

A one-on-one conversation works fine - we've been talking for ages without
incident. A third participant does not appear to cause problems either, but
when the fourth participant joins, people are starting to have client
crashes all the time (like once a minute or something). The interesting bit
is that this did not affect the user that was using Chrome 55 (who was the
first participant). The others were all using 58.

I've seen connectivity issues during "the Call" as well, as have I heard
of various references to browser bugs. Are my issues related? If so, what
exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?

Regards,

  Guus


#3

Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate some bugs (the ones within reach) but alas not all of them can be fixed from the client application. I’m afraid the only viable solution at the moment is not to use Chrome 58.

Cheers,

···

On Jun 1, 2017, at 11:58, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:

In my product, that wraps the videobridge, focus component and jitsi-meet, we're running into strange stability issues.

A one-on-one conversation works fine - we've been talking for ages without incident. A third participant does not appear to cause problems either, but when the fourth participant joins, people are starting to have client crashes all the time (like once a minute or something). The interesting bit is that this did not affect the user that was using Chrome 55 (who was the first participant). The others were all using 58.

I've seen connectivity issues during "the Call" as well, as have I heard of various references to browser bugs. Are my issues related? If so, what exactly are the details of the problem? Is the cause for this known? Anything to be done to mitigate the issue?

--
Saúl


#4

Is it known what browser characteristics are causing this? What's the
chance that these problems are resolved in 59? Any upstream bugs that we
can vote for?

···

On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> wrote:

> On Jun 1, 2017, at 11:58, Guus der Kinderen <guus.der.kinderen@gmail.com> > wrote:
>
> In my product, that wraps the videobridge, focus component and
jitsi-meet, we're running into strange stability issues.
>
> A one-on-one conversation works fine - we've been talking for ages
without incident. A third participant does not appear to cause problems
either, but when the fourth participant joins, people are starting to have
client crashes all the time (like once a minute or something). The
interesting bit is that this did not affect the user that was using Chrome
55 (who was the first participant). The others were all using 58.
>
> I've seen connectivity issues during "the Call" as well, as have I heard
of various references to browser bugs. Are my issues related? If so, what
exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?
>

Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate
some bugs (the ones within reach) but alas not all of them can be fixed
from the client application. I’m afraid the only viable solution at the
moment is not to use Chrome 58.

Cheers,

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev


#5

George will know for sure, but I think updating the bridge to the latest
and disabling lip-sync in meet's config should improve stability on Chrome
58.

Boris

···

On Thu, Jun 1, 2017 at 8:18 AM Guus der Kinderen < guus.der.kinderen@gmail.com> wrote:

Is it known what browser characteristics are causing this? What's the
chance that these problems are resolved in 59? Any upstream bugs that we
can vote for?

On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> > wrote:

> On Jun 1, 2017, at 11:58, Guus der Kinderen < >> guus.der.kinderen@gmail.com> wrote:
>
> In my product, that wraps the videobridge, focus component and
jitsi-meet, we're running into strange stability issues.
>
> A one-on-one conversation works fine - we've been talking for ages
without incident. A third participant does not appear to cause problems
either, but when the fourth participant joins, people are starting to have
client crashes all the time (like once a minute or something). The
interesting bit is that this did not affect the user that was using Chrome
55 (who was the first participant). The others were all using 58.
>
> I've seen connectivity issues during "the Call" as well, as have I
heard of various references to browser bugs. Are my issues related? If so,
what exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?
>

Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate
some bugs (the ones within reach) but alas not all of them can be fixed
from the client application. I’m afraid the only viable solution at the
moment is not to use Chrome 58.

Cheers,

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev


#6

I am running the latest bridge (as in, hours old), but that lip-sync
setting is something I can still do. Thanks!

- Guus

···

On 1 June 2017 at 15:45, Boris Grozev <boris@sip-communicator.org> wrote:

George will know for sure, but I think updating the bridge to the latest
and disabling lip-sync in meet's config should improve stability on Chrome
58.

Boris

On Thu, Jun 1, 2017 at 8:18 AM Guus der Kinderen < > guus.der.kinderen@gmail.com> wrote:

Is it known what browser characteristics are causing this? What's the
chance that these problems are resolved in 59? Any upstream bugs that we
can vote for?

On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> >> wrote:

> On Jun 1, 2017, at 11:58, Guus der Kinderen < >>> guus.der.kinderen@gmail.com> wrote:
>
> In my product, that wraps the videobridge, focus component and
jitsi-meet, we're running into strange stability issues.
>
> A one-on-one conversation works fine - we've been talking for ages
without incident. A third participant does not appear to cause problems
either, but when the fourth participant joins, people are starting to have
client crashes all the time (like once a minute or something). The
interesting bit is that this did not affect the user that was using Chrome
55 (who was the first participant). The others were all using 58.
>
> I've seen connectivity issues during "the Call" as well, as have I
heard of various references to browser bugs. Are my issues related? If so,
what exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?
>

Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate
some bugs (the ones within reach) but alas not all of them can be fixed
from the client application. I’m afraid the only viable solution at the
moment is not to use Chrome 58.

Cheers,

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev


#7

Is it known what browser characteristics are causing this? What's the chance that these problems are resolved in 59? Any upstream bugs that we can vote for?

Kinda, it’s a combination of things, and alas not all commits which fix them are public yet. I’ve been keeping track and I know 59 is fixed, but 58 is still affected, and I’m not sure if they will be back ported and a silent update will be pushed.

The areas which caused trouble are:

- simulcast
- stats
- some logging?

All our attempts at mitigating them all from our side have been futile, so we are in a “use Beta / Canary or wait” kinda situation, unfortunately.

Cheers,

···

On Jun 1, 2017, at 15:16, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:

On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> wrote:

> On Jun 1, 2017, at 11:58, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:
>
> In my product, that wraps the videobridge, focus component and jitsi-meet, we're running into strange stability issues.
>
> A one-on-one conversation works fine - we've been talking for ages without incident. A third participant does not appear to cause problems either, but when the fourth participant joins, people are starting to have client crashes all the time (like once a minute or something). The interesting bit is that this did not affect the user that was using Chrome 55 (who was the first participant). The others were all using 58.
>
> I've seen connectivity issues during "the Call" as well, as have I heard of various references to browser bugs. Are my issues related? If so, what exactly are the details of the problem? Is the cause for this known? Anything to be done to mitigate the issue?
>

Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate some bugs (the ones within reach) but alas not all of them can be fixed from the client application. I’m afraid the only viable solution at the moment is not to use Chrome 58.

Cheers,

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

--
Saúl


#8

Okay, thanks. Will enabling/disabling simulcast in the application then
make a difference? Stats, is that something that we can influence in the
bridge or in meet?

···

On 1 June 2017 at 17:37, Saúl Ibarra Corretgé <scorretge@atlassian.com> wrote:

> On Jun 1, 2017, at 15:16, Guus der Kinderen <guus.der.kinderen@gmail.com> > wrote:
>
> Is it known what browser characteristics are causing this? What's the
chance that these problems are resolved in 59? Any upstream bugs that we
can vote for?
>

Kinda, it’s a combination of things, and alas not all commits which fix
them are public yet. I’ve been keeping track and I know 59 is fixed, but 58
is still affected, and I’m not sure if they will be back ported and a
silent update will be pushed.

The areas which caused trouble are:

- simulcast
- stats
- some logging?

All our attempts at mitigating them all from our side have been futile, so
we are in a “use Beta / Canary or wait” kinda situation, unfortunately.

Cheers,

> On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> > wrote:
>
> > On Jun 1, 2017, at 11:58, Guus der Kinderen < > guus.der.kinderen@gmail.com> wrote:
> >
> > In my product, that wraps the videobridge, focus component and
jitsi-meet, we're running into strange stability issues.
> >
> > A one-on-one conversation works fine - we've been talking for ages
without incident. A third participant does not appear to cause problems
either, but when the fourth participant joins, people are starting to have
client crashes all the time (like once a minute or something). The
interesting bit is that this did not affect the user that was using Chrome
55 (who was the first participant). The others were all using 58.
> >
> > I've seen connectivity issues during "the Call" as well, as have I
heard of various references to browser bugs. Are my issues related? If so,
what exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?
> >
>
> Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate
some bugs (the ones within reach) but alas not all of them can be fixed
from the client application. I’m afraid the only viable solution at the
moment is not to use Chrome 58.
>
>
> Cheers,
>
> --
> Saúl
>
>
> _______________________________________________
> dev mailing list
> dev@jitsi.org
> Unsubscribe instructions and other list options:
> http://lists.jitsi.org/mailman/listinfo/dev
>
> _______________________________________________
> dev mailing list
> dev@jitsi.org
> Unsubscribe instructions and other list options:
> http://lists.jitsi.org/mailman/listinfo/dev

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev


#9

Okay, thanks. Will enabling/disabling simulcast in the application then make a difference? Stats, is that something that we can influence in the bridge or in meet?

The simulcast issue was work-arounded in the bridge. The stats crash wasn’t very clear, so I’m not sure where it stands. Sorry for being to vague, that’s pretty much all I know.

Cheers,

···

On Jun 1, 2017, at 17:56, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:

On 1 June 2017 at 17:37, Saúl Ibarra Corretgé <scorretge@atlassian.com> wrote:

> On Jun 1, 2017, at 15:16, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:
>
> Is it known what browser characteristics are causing this? What's the chance that these problems are resolved in 59? Any upstream bugs that we can vote for?
>

Kinda, it’s a combination of things, and alas not all commits which fix them are public yet. I’ve been keeping track and I know 59 is fixed, but 58 is still affected, and I’m not sure if they will be back ported and a silent update will be pushed.

The areas which caused trouble are:

- simulcast
- stats
- some logging?

All our attempts at mitigating them all from our side have been futile, so we are in a “use Beta / Canary or wait” kinda situation, unfortunately.

Cheers,

> On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> wrote:
>
> > On Jun 1, 2017, at 11:58, Guus der Kinderen <guus.der.kinderen@gmail.com> wrote:
> >
> > In my product, that wraps the videobridge, focus component and jitsi-meet, we're running into strange stability issues.
> >
> > A one-on-one conversation works fine - we've been talking for ages without incident. A third participant does not appear to cause problems either, but when the fourth participant joins, people are starting to have client crashes all the time (like once a minute or something). The interesting bit is that this did not affect the user that was using Chrome 55 (who was the first participant). The others were all using 58.
> >
> > I've seen connectivity issues during "the Call" as well, as have I heard of various references to browser bugs. Are my issues related? If so, what exactly are the details of the problem? Is the cause for this known? Anything to be done to mitigate the issue?
> >
>
> Yes, it looks related. Chrome 58 is a “bad apple”. We tried to mitigate some bugs (the ones within reach) but alas not all of them can be fixed from the client application. I’m afraid the only viable solution at the moment is not to use Chrome 58.
>
>
> Cheers,
>
> --
> Saúl
>
>
> _______________________________________________
> dev mailing list
> dev@jitsi.org
> Unsubscribe instructions and other list options:
> http://lists.jitsi.org/mailman/listinfo/dev
>
> _______________________________________________
> dev mailing list
> dev@jitsi.org
> Unsubscribe instructions and other list options:
> http://lists.jitsi.org/mailman/listinfo/dev

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev

--
Saúl


#10

Now that Chrome 59 is available, most of the stability issues that we had
have disappeared.

···

On 2 June 2017 at 09:56, Saúl Ibarra Corretgé <scorretge@atlassian.com> wrote:

> On Jun 1, 2017, at 17:56, Guus der Kinderen <guus.der.kinderen@gmail.com> > wrote:
>
> Okay, thanks. Will enabling/disabling simulcast in the application then
make a difference? Stats, is that something that we can influence in the
bridge or in meet?
>

The simulcast issue was work-arounded in the bridge. The stats crash
wasn’t very clear, so I’m not sure where it stands. Sorry for being to
vague, that’s pretty much all I know.

Cheers,

> On 1 June 2017 at 17:37, Saúl Ibarra Corretgé <scorretge@atlassian.com> > wrote:
>
> > On Jun 1, 2017, at 15:16, Guus der Kinderen < > guus.der.kinderen@gmail.com> wrote:
> >
> > Is it known what browser characteristics are causing this? What's the
chance that these problems are resolved in 59? Any upstream bugs that we
can vote for?
> >
>
> Kinda, it’s a combination of things, and alas not all commits which fix
them are public yet. I’ve been keeping track and I know 59 is fixed, but 58
is still affected, and I’m not sure if they will be back ported and a
silent update will be pushed.
>
> The areas which caused trouble are:
>
> - simulcast
> - stats
> - some logging?
>
> All our attempts at mitigating them all from our side have been futile,
so we are in a “use Beta / Canary or wait” kinda situation, unfortunately.
>
>
> Cheers,
>
> > On 1 June 2017 at 15:06, Saúl Ibarra Corretgé <scorretge@atlassian.com> > wrote:
> >
> > > On Jun 1, 2017, at 11:58, Guus der Kinderen < > guus.der.kinderen@gmail.com> wrote:
> > >
> > > In my product, that wraps the videobridge, focus component and
jitsi-meet, we're running into strange stability issues.
> > >
> > > A one-on-one conversation works fine - we've been talking for ages
without incident. A third participant does not appear to cause problems
either, but when the fourth participant joins, people are starting to have
client crashes all the time (like once a minute or something). The
interesting bit is that this did not affect the user that was using Chrome
55 (who was the first participant). The others were all using 58.
> > >
> > > I've seen connectivity issues during "the Call" as well, as have I
heard of various references to browser bugs. Are my issues related? If so,
what exactly are the details of the problem? Is the cause for this known?
Anything to be done to mitigate the issue?
> > >
> >
> > Yes, it looks related. Chrome 58 is a “bad apple”. We tried to
mitigate some bugs (the ones within reach) but alas not all of them can be
fixed from the client application. I’m afraid the only viable solution at
the moment is not to use Chrome 58.
> >
> >
> > Cheers,
> >
> > --
> > Saúl
> >
> >
> > _______________________________________________
> > dev mailing list
> > dev@jitsi.org
> > Unsubscribe instructions and other list options:
> > http://lists.jitsi.org/mailman/listinfo/dev
> >
> > _______________________________________________
> > dev mailing list
> > dev@jitsi.org
> > Unsubscribe instructions and other list options:
> > http://lists.jitsi.org/mailman/listinfo/dev
>
> --
> Saúl
>
>
> _______________________________________________
> dev mailing list
> dev@jitsi.org
> Unsubscribe instructions and other list options:
> http://lists.jitsi.org/mailman/listinfo/dev
>
> _______________________________________________
> dev mailing list
> dev@jitsi.org
> Unsubscribe instructions and other list options:
> http://lists.jitsi.org/mailman/listinfo/dev

--
Saúl

_______________________________________________
dev mailing list
dev@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/dev