Hello Alex!
I dont think you can handle such many calls with gnugk. I'm using it at
a dual xeon 3.4 GHz and it can handle 130-170 simultanous calls at
maximum. However, only a single CPU is used, because with FreeBSD 5.4
gnugk can handle only very few calls (10-30) with a SMP kernel. If you
assume an average call duration of only 120 that are not more than 1-2
new calls per second in average.
Best regards
Jan
Teodor Georgiev schrieb:
Try to increase the call signalling handler value and tell if there is
any improvement.
----- Original Message ----- From: "Alex Rubenstein" <alex@xxxxxxx>
To: <openh323gk-users@xxxxxxxxxxxxxxxxxxxxx>
Sent: Saturday, July 16, 2005 6:40 AM
Subject: Help! Performance issues..
(all of the OS and information about the GK is at the end of the email).
Hello,
The machine is a dual xeon 2.4 ghz, with hyperthreading disabled.
I am running gnugk 2.2.2, which is gatekeeping for about 3 or 4 cisco
gateways, and a couple Cisco CCMs. 99% of the time, everything is
A-OK. 99% of the time, the system is not busy -- perhaps 1 to 10
calls per minute.
However, we have a application that triggers a huge amount of inbound
calls, perhaps 10 to 30 per second, several times per week. It will
largely fill up several ISDN-PRIs for a period of about 10 to 25
minutes.
During this time, the gatekeeper does not crash; however, it does
seem to be entirely consumed handling setups and teardowns (many,
many of these calls are rejected due to capacity issues on the
terminated gateway).
The config file running on the GK is pretty vanilla, with the
exception that there are about 1,950 RewriteE164 lines.
During this time CPU is pretty pegged at about 60% (which is across
both processors). I am making the assumption that gnugk is taking a
non-threaded approach and killing one CPU, and the other is doing god
knows what.
So, my questions:
a) should gnugk kill/bring to it's knees a dual xeon 2.4 with 10 to
30 setups/teardowns per second?
b) Am I missing something planly obvious?
The GK normally sits around at about 0 to 3% CPU with the above 99%
normal low load. However, when I sit at my desk and redial/hangup
from an ip phone 2 times a second, CPU rides up to about 8%. Does
this sound right? Seems that it shouldn't be _that_ busy doing this.
Any help appreciated.
--------------------------------------------
#
# gnugk.ini, made by gk-ini.pl, Mon Jun 27 21:34:22 2005
#
[Gatekeeper::Main]
Fourtytwo=42
Name=zz1.yyy
TimeToLive=300
[RasSrv::RRQFeatures]
AcceptGatewayPrefixes=1
[RasSrv::PermanentEndpoints]
10.10.10.10=dead;99999#
[RoutedMode]
GKRouted=1
H245Routed=0
RemoveH245AddressOnTunneling=0
SendReleaseCompleteOnDRQ=0
[Endpoint]
TimeToLive=900
[RasSrv::RewriteE164]
800zzz2920=10002#800zzz2920
800zzz3384=10002#800zzz3384
800zzz3404=10002#800zzz3404
800zzz0328=10002#800zzz0328
[...]
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________________
Posting: mailto:Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549
Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users
Homepage: http://www.gnugk.org/
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________________
Posting: mailto:Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549
Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users
Homepage: http://www.gnugk.org/
