On 02/18/2011 12:53 PM, Thomas Baumann wrote:
Hello Marius,
the main problem I had was caused with wrong debug-level in the
script. Debug level 3 is not a good choice for loadtesting, my
mistake.
After changing to debug level 2, it looks a lot better. I did more
testing with the right settings, and yes the change in pdb.c had some
influence on the timeouts.
Hello Thomas,
Good to hear that, I couldn't replicate the problems you were having
with pdb. Are you sure you have asynchronous syslogd configured for
kamailio? If not performance is a killer
BTW, my patch doesn't really do anything afaik (i am not 100% sure on
linux but my tests seemed to confirm it).
Cheers,
Marius
With the patch I did not see any timeouts after running a loadtest
overnight (200.000 calls executed with max. 12 cps for one application
server). But with the old code where are still few timeouts:
Completed Calls:9732 Failed Calls:13 Failure Rate:0.133 perc. Loop Count:2336
Have a nice weekend,
Thomas
On Wed, Feb 16, 2011 at 8:20 PM, bt047265<bt047...@gmail.com> wrote:
Hi Marius, Hi Henning,
sorry for the wrong information, the patch did not work for me. I did
not see any timeouts if only one server was used. If more than 2
Servers are used, where are failed calls.
regards,
Thomas
One server old pdb.c:
modparam("pdb", "server", "10.12.18.41:10001")
Feb 16 19:28:03 m-lab-prx-as-2 /openser/sbin/kamailio[29622]: INFO:
pdb [pdb.c:291]: got an answer in 0.990000 ms
Feb 16 19:28:03 m-lab-prx-as-2 /openser/sbin/kamailio[29619]: INFO:
pdb [pdb.c:291]: got an answer in 36.197000 ms
Feb 16 19:28:03 m-lab-prx-as-2 /openser/sbin/kamailio[29619]: INFO:
pdb [pdb.c:291]: got an answer in 0.151000 ms
two servers old pdb.c:
modparam("pdb", "server", "10.12.18.41:10001,10.12.19.51:10001")
Feb 16 19:35:13 m-lab-prx-as-2 /openser/sbin/kamailio[29688]: INFO:
pdb [pdb.c:291]: got an answer in 0.179000 ms
Feb 16 19:35:13 m-lab-prx-as-2 /openser/sbin/kamailio[29693]: INFO:
pdb [pdb.c:291]: got an answer in 0.248000 ms
Feb 16 19:35:13 m-lab-prx-as-2 /openser/sbin/kamailio[29692]: WARNING:
pdb [pdb.c:260]: exceeded timeout while waiting for response.
Feb 16 19:35:13 m-lab-prx-as-2 /openser/sbin/kamailio[29688]: INFO:
pdb [pdb.c:291]: got an answer in 0.489000 ms
Feb 16 19:35:13 m-lab-prx-as-2 /openser/sbin/kamailio[29692]: INFO:
pdb [pdb.c:291]: got an answer in 0.596000 ms
One server new pdb.c:
modparam("pdb", "server", "10.12.18.41:10001")
Feb 16 19:55:31 m-lab-prx-as-2 /openser/sbin/kamailio[1079]: INFO: pdb
[pdb.c:291]: got an answer in 48.584000 ms
Feb 16 19:55:31 m-lab-prx-as-2 /openser/sbin/kamailio[1084]: INFO: pdb
[pdb.c:291]: got an answer in 0.290000 ms
Feb 16 19:55:31 m-lab-prx-as-2 /openser/sbin/kamailio[1079]: INFO: pdb
[pdb.c:291]: got an answer in 0.190000 ms
Feb 16 19:55:31 m-lab-prx-as-2 /openser/sbin/kamailio[1079]: INFO: pdb
[pdb.c:291]: got an answer in 0.241000 ms
two servers old pdb.c:
modparam("pdb", "server", "10.12.18.41:10001,10.12.19.51:10001")
Feb 16 20:08:36 m-lab-prx-as-1 /openser/sbin/kamailio[17348]: INFO:
pdb [pdb.c:291]: got an answer in 0.314000 ms
Feb 16 20:08:36 m-lab-prx-as-1 /openser/sbin/kamailio[17348]: INFO:
pdb [pdb.c:291]: got an answer in 0.202000 ms
Feb 16 20:08:36 m-lab-prx-as-1 /openser/sbin/kamailio[17349]: WARNING:
pdb [pdb.c:260]: exceeded timeout while waiting for response.
- Show quoted text -
On Wed, Feb 16, 2011 at 4:59 PM, marius zbihlei<marius.zbih...@1and1.ro> wrote:
On 02/16/2011 04:47 PM, Thomas Baumann wrote:
Hello Marius,
thanks a lot for submitting the patch. It works like a charm now. But to
be honest I had troubles to find this patch, I just checked the daily
tarball [http://sip-router.org/tarballs/sr/] and found this change.
How I can get patch e914f97b4 in a better way ?
regards,
Hello
You can find here a tutorial on installing kamailio from GIT.
http://www.kamailio.org/dokuwiki/doku.php/install:kamailio-3.1.x-from-git
Glad the patch worked. I really doubted the usefulness of the patch as
revents member should be cleared by the poll() method before setting the
POLLIN event for a fd. I will have to test it with a small test case on my
machine before I commit this to master branch(anyway no wrong can come to
it, the change has no impact on performance and stability of the pdb module)
Cheers
Marius
Thomas
Hello
I have made a patch to 3.1(e914f97b4) . The poll() event list was not
cleared when a response was received so when 2 servers responded one
after the other it might block in an incorrect recv(). Can you please
test with this?
Marius
___________________________________________________________
Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die
Toolbar eingebaut! http://produkte.web.de/go/toolbar
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users
