We had our first big issues with our freeswitch system today. During at
least 2 conferences, audio became jittery and there were three occasions
where everyone was dropped from a conference. Even so, conference recording
was not interrupted, and the freeswitch debug log doesn't show anything
unusua
Stephen Crosby wrote:
> We had our first big issues with our freeswitch system today. During at
> least 2 conferences, audio became jittery and there were three occasions
> where everyone was dropped from a conference. Even so, conference recording
> was not interrupted, and the freeswitch debug l
To answer the other question, I did not try pinging the host while the
problems were occurring, but I was able to call back in with no problem only
a few seconds after I got dropped (less than 20), so if it is a network
issue it would have to have been very brief.
--Stephen
On Mon, May 4, 2009 at
Network problem is what I'm still thinking. Take a look at this log snippet:
http://pastebin.freeswitch.org/8813
I'm CALLER_A, and you can see me calling back in on line 8 after I got
dropped. But that was only seconds prior. There really seems to be nothing
in the log at the time the calls were
Sorry for the extra messages, but I've just discovered something that
probably helps pinpoint the problem:
http://pastebin.freeswitch.org/8814
It seems that the callers were disconnected, but freeswitch had to wait a
timeout period before it actually hangs up which looks like about 5 minutes.
So I
Stephen Crosby wrote:
> Network problem is what I'm still thinking. Take a look at this log snippet:
>
> http://pastebin.freeswitch.org/8813
>
> I'm CALLER_A, and you can see me calling back in on line 8 after I got
> dropped. But that was only seconds prior. There really seems to be nothing
> i
Stephen Crosby wrote:
> Sorry for the extra messages, but I've just discovered something that
> probably helps pinpoint the problem:
> http://pastebin.freeswitch.org/8814
>
> It seems that the callers were disconnected, but freeswitch had to wait a
> timeout period before it actually hangs up whi
I was on this conference and there were 4 of us that got dropped at the same
time. By the way, these were all regular PSTN lines calling through our SIP
provider and being routed to our freeswitch instance which is being hosted
on a VPS.
When you say a device on my end terminated the call, do you
Stephen Crosby wrote:
> I was on this conference and there were 4 of us that got dropped at the same
> time. By the way, these were all regular PSTN lines calling through our SIP
> provider and being routed to our freeswitch instance which is being hosted
> on a VPS.
>
> When you say a device on
First off you're not on SVN trunk secondly Are you executing the
conference app inside your js file? If so then there could be the
problem! You have also forgotten to include anything about Distro,
OS, CPU and Memory.
/b
On May 5, 2009, at 12:12 AM, Stephen Crosby wrote:
We had our firs
I know I'm not on svn trunk, but this is a production server and it's just
not feasible to update it constantly. I can update it though if you think I
need to. I am routing callers to the conference app with javascript like
this:
session.execute("conference", xyz);
Can you tell me more about the pr
Forgot to add that my OS is Ubuntu 8.04LTS (hardy heron).
--Stephen
On Tue, May 5, 2009 at 7:42 AM, Stephen Crosby wrote:
> I know I'm not on svn trunk, but this is a production server and it's just
> not feasible to update it constantly. I can update it though if you think I
> need to. I am ro
You should rule out the network problems first, which sound more likely.
you can reduce the overuse of JS if you transfer the call to a regular
extension with a dynamic regex.
session.execute("transfer", "conf-xyz");
then make a regex in your xml dialplan to pick up ^conf-(.*) and execute
confer
Why is initiating a conference from a dial plan entry better than
directly from an internal script? Is there a page that details what
is and isn't script abuse?
On 5/5/09, Anthony Minessale wrote:
> You should rule out the network problems first, which sound more likely.
>
> you can reduce the o
Javascript has it's own issues. Tony wrote FreeSWITCH, including the
dialplan parser. The less you can use JS, especially in a high-volume
environment, the better. The happy medium is to use Lua. It is small,
lightweight, easy, and designed to be embedded. Tony has had 3000 Lua
session up simultane
The script has a javascript interpreter allocated sitting there eating
up ram for no reason... so to scale better its wise to exit the script
and do your conference.
Lua is lighter weight but still having the interpreter hang around
when not needed isn't wise.
/b
On May 14, 2009, at 11:3
I keep hearing that lua is lighter weight / more scalable than javascript.
I'd love to see some data that shows how big the difference really is. I
could port all my scripts from javascript to lua, but it would require a lot
of overhead (me learning lua + actually porting the scripts). I want to kn
Its obvious if you look at the size of the JS VM vs the lua VM.. it
would clearly scale better not needing megs and megs of ram per call
vs about 160kb for lua. Lua took about 20 min to learn and about an
hour to get the finer points down.
/b
On May 14, 2009, at 12:24 PM, Stephen Crosby
How would using the mod_event_socket compare to lua / javascript with
regards performance and scaling? (ignoring the client side of the
socket, just the Freeswitch side). Is there any performance issues to
placing a call into conference using with callflow from the
mod_socket?
N.
2009/5/14 Brian
I think its less obvious to a user who'd never heard of lua until recently.
I think it would be very helpful if someone had a comparison and could put
it on the wiki.
I spent some time poking around on the wiki before I decided to write my
scripts in javascript. If I had seen any warnings on
scala
handy benefit for LUA, you can configure LUA scripts to run at switch startup to perform tasks (like a cron system) within the switch core. -pete
Original Message
Subject: Re: [Freeswitch-users] help with mod_conference stability
From: Brian West
Date: Thu, May 14, 2009 10:31
tasks (like a cron system) within the switch
core.
>
> -pete
>
> ---- Original Message ----
> Subject: Re: [Freeswitch-users] help with mod_conference stability
> From: Brian West
> Date: Thu, May 14, 2009 10:31 am
> To: freeswitch-users@lists.freeswitch.org
>
> It
On Thu, 2009-05-14 at 12:31 -0500, Brian West wrote:
> Its obvious if you look at the size of the JS VM vs the lua VM.. it
> would clearly scale better not needing megs and megs of ram per call
> vs about 160kb for lua.
Isn't this a bit of a non sequitur, given that there ought to be just
one cop
Unfortunately, no scripting language we have encountered
has ever done proper threading to the point that you could even begin to do
that.
Lua is one of the better ones in a threaded env only because they solved the
problem by making a tiny VM that
can be easily reproduced thousands of times. You
24 matches
Mail list logo
