VoIP Mailing List Archives

[pmhshz at gmail.com]

Hi Brian,

I am first looking to modify the script, as there are several things to
modify (thanks to Anthony) then paralally I will test 1.0.3 RC1 also and any
new thing come out then I will post update here...

Thanks,
msp

Brian West-3 wrote:

[brian at freeswitch.org]

The RC1 tarball has all the SVN dirs so you can "make current" on it.

/b

On Feb 1, 2009, at 11:26 PM, shehzad p wrote:

[pmhshz at gmail.com]

Hi anthony,

I Modified the whole architecture of call routing system,
Now after getting required routes, script exit and,
control comes back to Dialplan, and call is bridged there,
And call hangup, CDR is posted to cdr.php file (using xml_cdr).

So now there is no blocking statement (bridge or anything like that) in
current javascript, It return back control instantly.

So, setting up all above architecture...
First I tested FS 1.0.1 , It get crashed two times, in interval of 3 to 5
hours and simultaneous call of about 100 to 150.
BT is the same as before...
http://www.nabble.com/file/p21825226/bt_new_arch.txt bt_new_arch.txt

Now I am also testing 1.0.3RC1, and post it back if any found.

Thanks
msp


Clearly you have an issue with your javascript code.

You have the Garbage collector blocking in every thread.

Are you doing any endless loops in your code where you do not check
session.ready() as a condition for
continuing the script?

any time session.ready() fails you must immediately exit.

Are you using session.execute to execute long blocking operations like
bridging many calls or entering a conference?
You should avoid doing this as all the collective scripts on the system
share a common Garbage Collector provided by the
JS engine and it can lead to the exact issues you describe if the code is
not properly designed.

What else does you script do that are things provided by FS such as playing
files and executing applications.




--
View this message in context: http://www.nabble.com/Freeswitch-freezes-on-increasing-call-traffic-tp21701744p21825226.html
Sent from the Freeswitch-users mailing list archive at Nabble.com.


_______________________________________________
Freeswitch-users mailing list
Freeswitch-users@lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

[anthony.minessale at g...]

If you still get a crash on SVN trunk please post the bt even if you think it's the same, since it won't be
exactly the same, the line numbers etc will be accurate with our development code making it easier to debug.


On Wed, Feb 4, 2009 at 12:38 AM, shehzad p <pmhshz@gmail.com (pmhshz@gmail.com)> wrote:

[pmhshz at gmail.com]

Hi anthony,
In my previous post I already attached the BT for the testing of FS 1.0.1
Posting it again, please find it on
link==>http://www.nabble.com/file/p21825226/bt_new_arch.txt

Now I got the same result while testing FS 1.0.3RC1, And its BT is also
same... BT Link:
http://www.nabble.com/file/p21847332/fs_1_0_3_bt_new_arch.log
fs_1_0_3_bt_new_arch.log

(Note: Same in the sens the functions listed in the sequence are almost same
as before...)


Anthony Minessale-2 wrote:

[brian at freeswitch.org]

Can you give me the output of uname -a and the contents of /proc/
cpuinfo? Not sure I asked for this info already or not.

Thanks,
Brian

On Feb 5, 2009, at 2:42 AM, shehzad p wrote:

[pmhshz at gmail.com]

HI Brian,

Output of ulimit -a and /proc/cpuinfo is attached...
http://www.nabble.com/file/p21848148/12_ulimit_and_cpuinfo.log
12_ulimit_and_cpuinfo.log

BUT......................
I am running the freeswitch using below command (So ulimit set according to
Anthony's previous post):
===================================================================================
ulimit -d unlimited; ulimit -f unlimited; ulimit -i unlimited; ulimit -n
999999; ulimit -q unlimited ; ulimit -u unlimited; ulimit -v unlimited ;
ulimit -x unlimited; ulimit -s 244; ulimit -l unlimited; freeswitch
===================================================================================

Thanks
msp



Brian West-3 wrote:

[pmhshz at gmail.com]

Hi Brian,

As it can be seen from the system information, there require any change in
system or any suggestion...

out put of uname -a is :
Linux localhost.localdomain 2.6.23.1-42.fc8 #1 SMP Tue Oct 30 13:55:12 EDT
2007 i686 i686 i386 GNU/Linux


Thanks,
msp


shehzad p wrote:

[anthony.minessale at g...]

First of all please stop using the mailing list as a bug tracker.
All issues should be put into jira and managed with that.

Secondly, Didn't I ask you multiple times to stop using release snapshots and please use the SVN
trunk? I don't understand why you keep ignoring me and using everything but what I asked.

I am not telling you to use SVN because I think it will be fixed it's so we are on the development copy of the code
to get the proper line numbers etc. If you look at your 2 bt you posted, the line numbers are different on each one.

What are you using on the other side of ODBC? as you can see in your bt, the call goes into ODBC then into several
libs with no symbols and crashes on free. This can be a sign of corrupt memory, running out of memory or an issue in either ODBC or the database specific lib.

What distro is it?
What ODBC version? unixODBC? version xxx?
What database driver version xxx? Is it mysl not using the proper reentrant version of the plugin?
Sometimes packaged libs have bugs in them which fall out of our control.
Can you build unixODBC and the plugins yourself with debug symbols so we can see if that is the cause or at the very least then we can see the debug info in the bt.

please make sure you address *all* my questions in your jira report. Starting with using svn trunk, *hint* type "make current" from your rc1 distro.




On Thu, Feb 5, 2009 at 3:51 AM, shehzad p <pmhshz@gmail.com (pmhshz@gmail.com)> wrote:

[paul.degt at gmail.com]

Look like you use Fedora. I had a lot of issues with using Fedora as
production or load test system, in my opinion it's more like work in
progress than a production ready stable linux. If you cannot buy RHEL or
SLES use Centos.

shehzad p wrote:

[pmhshz at gmail.com]

Hi all,

I have opened JIRA for the same.
http://jira.freeswitch.org/browse/FSCORE-285

One system is Fedora, and another one is Ubuntu. Although fs 1.0.1 was also
get crashed in Ubuntu many times.
Now 1.0.3.RC1 is loaded on Ubuntu, so this happen again with my Ubuntu I
will surely post it.

And I am now going to test latest trunk version and post back if any thing
found...

thanks,
msp

paul.degt wrote:

[anthony.minessale at g...]

I am providing you with free software and free support.
I am willing to investigate you problem for you but you must be more cooperative.

I took the time to ask you several questions yesterday and you have sent 4 emails since and not addressed it.

Please move any further correspondence on this issue to your jira ticket.

If you do not have the necessary skill to build your own ODBC and driver as the email requests,

will you please try CentOS 64 bit linux with only SVN Trunk which is a platform we trust the unixODBC and its depends.

Please E-mail me privately your latest JS code.

Also please try to produce a minimal script that will reproduce your issue that we can reproduce.

If you continue to ignore my requests I will have no choice but to close your issue.

Transcript of ignored e-mail
----------------------------------

First of all please stop using the mailing list as a bug tracker.
All issues should be put into jira and managed with that.

Secondly, Didn't I ask you multiple times to stop using release snapshots and please use the SVN
trunk? I don't understand why you keep ignoring me and using everything but what I asked.

I am not telling you to use SVN because I think it will be fixed it's so we are on the development copy of the code
to get the proper line numbers etc. If you look at your 2 bt you posted, the line numbers are different on each one.

What are you using on the other side of ODBC? as you can see in your bt, the call goes into ODBC then into several
libs with no symbols and crashes on free. This can be a sign of corrupt memory, running out of memory or an issue in either ODBC or the database specific lib.

What distro is it?
What ODBC version? unixODBC? version xxx?
What database driver version xxx? Is it mysl not using the proper reentrant version of the plugin?
Sometimes packaged libs have bugs in them which fall out of our control.
Can you build unixODBC and the plugins yourself with debug symbols so we can see if that is the cause or at the very least then we can see the debug info in the bt.

please make sure you address *all* my questions in your jira report. Starting with using svn trunk, *hint* type "make current" from your rc1 distro.


On Fri, Feb 6, 2009 at 2:30 AM, shehzad p <pmhshz@gmail.com (pmhshz@gmail.com)> wrote: