[jitsi-users] GUI freezing in Linux and Windowze crash


#1

As we deployed Jitsi to a larger number of volunteer users, we started receiving reports of two kinds of issues.

One is that in Linux user interface would kind of freeze. I am saying kind of because it just happens when user receives or starts more than one chat session and tries to switch between chat window tabs. They don't switch and since that attempt GUI elements seem to be frozen: cannot use drop down boxes in account settings and cannot press the buttons. Window can only be closed with title X button.

Another issue is that sometimes random users receive a chat message and immediately their computers crash. We observed that they crash when someone sends them a very first chat message and before chat window opens. It is weird as the very same 2 users can chat all day long and then boom and another message crashes. Only happens on Windows platform.

Any suggestions?

Still waiting patiently for recommendations about audio calls not working :frowning:

figoli


#2

Could you please send us the logs (https://jitsi.org/logs) of the affected clients?

And if you say the computer crash, do you mean the actual operatin system? If so, then there is most probably something wrong with the drivers of the sound card(s).

Freundliche Grüsse,
Ingo Bauersachs

-- sent from my mobile

···

Le 23.01.2014 à 00:17, "info" <info@dressmaker.ca> a écrit :

As we deployed Jitsi to a larger number of volunteer users, we started receiving reports of two kinds of issues.

One is that in Linux user interface would kind of freeze. I am saying kind of because it just happens when user receives or starts more than one chat session and tries to switch between chat window tabs. They don't switch and since that attempt GUI elements seem to be frozen: cannot use drop down boxes in account settings and cannot press the buttons. Window can only be closed with title X button.

Another issue is that sometimes random users receive a chat message and immediately their computers crash. We observed that they crash when someone sends them a very first chat message and before chat window opens. It is weird as the very same 2 users can chat all day long and then boom and another message crashes. Only happens on Windows platform.

Any suggestions?

Still waiting patiently for recommendations about audio calls not working :frowning:

figoli

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


#3

Ingo Bauersachs <ingo@jitsi.org> wrote :

> Could you please send us the logs
> (https://jitsi.org/logs) of the
> affected clients? And if you say

Yes, we can request the user to ship the logs to us and once received, we'll
forward. In the mean time I will build a VM with their OS here and try to
reproduce.

> the computer crash, do you mean
> the actual operatin system? If so,
> then there is most probably
> something wrong with the drivers
> of the sound card(s).

You mean somehow incoming message uses sound driver in a manner that can crash a
computer's OS? If that was the case, we would be observing crashes on those
machines all day long. However nothing else but Jitsi incoming chat crashes it.
We can try turning off incoming message sound and give it a few days, but I
would be dumbfounded if that was the root cause. Don't get me wrong: these PCs
are under very heavy use and not just used as typewriters.

···

___________________________________
NOCC, http://nocc.sourceforge.net


#4

the computer crash, do you mean
the actual operatin system? If so,
then there is most probably
something wrong with the drivers
of the sound card(s).

You mean somehow incoming message uses sound driver in a manner that can
crash a computer's OS? If that was the case, we would be observing
crashes on those machines all day long. However nothing else but Jitsi
incoming chat crashes it. We can try turning off incoming message sound
and give it a few days, but I would be dumbfounded if that was the root
cause. Don't get me wrong: these PCs are under very heavy use and not
just used as typewriters.

The point I'm trying to make is: Jitsi is a user-mode application. If the
system crashes (and not just the Jitsi) then it must somehow be related to a
driver or the kernel itself. It might well be possible that Jitsi accesses
the sound card in a way no other application does and thus is the only
application causing the bug.
But without more details and logs this is really guesswork.

Ingo


#5

Please write to the list and not to me directly.

The point I'm trying to make is: Jitsi is a user-mode application. If
the system crashes (and not just the Jitsi) then it must somehow be
related to a driver or the kernel itself. It might well be possible
that Jitsi accesses the sound card in a way no other application does
and thus is the only application causing the bug. But without more
details and logs this is really guesswork.

Ingo

I would rather blame Java than Jitsi. We are observing some pretty
horrible things with Java in our development and created countless
issues to Oracle which is like a black box from that perspective. But
you are right and I have no idea what you are doing in your code with
audio. For example, are you using any custom native DLLs for audio/video
IO?

Yes, most of this is done natively. However that doesn't change the fact that a user-mode application is unable to crash the OS unless there is a fault in a driver or the kernel.

figoli

Ingo


#6

Yes, most of this is done natively. However that doesn't change the fact that
a user-mode application is unable to crash the OS unless there is a fault in a
driver or the kernel.

> figoli

Ingo

Agree. However when it is just one application out of a dozen more that work
fine, it's still an application issue, as the driver is unlikely to change.
Perhaps you can argue that IBM, Intel or Realtek have bugs in their driver, but
when yours is the only program that traps the OS, than you don't have a leg to
stand on and you might want to review the native libs you are using.

figoli


#7

Ingo:

You should set up your email like the rest of the contributors so that it sends on behalf of users@ to make sure emails go to the list. It works for everyone here, but you, and only but some of your emails. I don't know why all of a sudden your next email arrives from your personal address and not from the list. I don't have brains to check each and every sender's address, sorry.

Anyhow, I just reproduced the freezing of UI here. Opened chat with one user, than with another and when I went to authenticate the first user, UI is frozen and I cannot switch authentication type from question to fingerprint. This is in Fedora 19 with 3.12 kernel and under KDE 4.11.5

I was already going to send the logs but noticed that you are exporting too much personally identifiable information and I don't have authorization from those users to disclose that, so first I will have to set up the test accounts without real names and emails and only then I'll be able to send the logs. Something to think about, like masking email addresses during the export.

figoli

···

On 01/23/2014 12:50 PM, Ingo Bauersachs wrote:

Please write to the list and not to me directly.

The point I'm trying to make is: Jitsi is a user-mode application. If
the system crashes (and not just the Jitsi) then it must somehow be
related to a driver or the kernel itself. It might well be possible
that Jitsi accesses the sound card in a way no other application does
and thus is the only application causing the bug. But without more
details and logs this is really guesswork.

Ingo

I would rather blame Java than Jitsi. We are observing some pretty
horrible things with Java in our development and created countless
issues to Oracle which is like a black box from that perspective. But
you are right and I have no idea what you are doing in your code with
audio. For example, are you using any custom native DLLs for audio/video
IO?

Yes, most of this is done natively. However that doesn't change the fact that a user-mode application is unable to crash the OS unless there is a fault in a driver or the kernel.
  

figoli

Ingo