Bogdan (and list),
Again, thank you very much for taking the time to answer my questions. It
is people like you that enable the open source community to thrive and
grow. I really appreciate your efforts.
I have gone through and made your suggested changes and things seem to be
working quite well for what I am trying to accomplish.
My last and final question to the group (for now ;)... how do you determine
the dimensions of an OpenSIPS deployment? I am going to be running two of
these systems with automatic failover. They will be running on Dell 1950
with 8 cores of CPU (2.0Ghz) and 8G or RAM on CentOS x_86/64. OpenSIPS is
the only extra process running on the systems. Will this configuration
handle 1,000 concurrent calls, 10,000 concurrent calls, etc? Will my limits
be in Call Setups Per Second? How does one go about estimating the capacity
of OpenSIPS?
Thanks,
Geoff
On Nov 13, 2008 5:02am, Bogdan-Andrei Iancu <[EMAIL PROTECTED]> wrote:
Hi Geoff,
[EMAIL PROTECTED] wrote:
Thank you very much for taking the time to look over my configuration. I
just want to make sure of something. I replied to my own original message
with a greatly enhanced configuration. I realized the first was missing a
huge amount of logic after studying up on OpenSIPS for 2 days. Were you
commenting on the original message, or the second message? Based on my
testing, i experienced slightly different results than you described.
My comments were generally speaking (for a LB topic) and not strictly in
regards to a particular script.
What I am seeing (based on the second config) is that only the initial
INVITE falls into the route(1) block, which is the way I intended it. This
means only the INVITE requests are routed via the ds_select_dst() call to
the dispatcher. All subsequent messages fall into my loose_route() check
and are simply relayed via t_relay().
yes, because you still so record_route() - if you remove this, you will
process only the initial INVITEs - the ACK, BYEs will not even pass thorugh
your server.
I included a little snippet of the logging I do via the xlog calls. One
thing that confuses me is that based on my configuration and my logs, I
never explicitly relay the TRYING or OK messages. I set up my
onreply_route[1], but all I do is log that I got the reply. I did this
because regardless of what I do here, the UAC which requested the INVITE
gets the TRYING and OK messages properly. Is there something in the tm.so
that implicitly handles these, or am I missing some big picture element
here.
replies are automatically routed back on the reverted path of the request
- you do not need to route them explicitly.
Regards,
Bogdan
Thanks!
Geoff
################ BEGIN LOG SNIPPET ########################
New request - M=INVITE RURI=sip:[EMAIL PROTECTED] F=sip:[REMOVED]
T=sip:[EMAIL PROTECTED] IP=10.2.252.190
[EMAIL PROTECTED]
Recording Route info
Method is an INVITE, fetching next from dispatcher
Reply - S=100 D=Trying F=sip:[REMOVED] T=sip:[EMAIL PROTECTED]
IP=10.2.252.181 [EMAIL PROTECTED]
Reply - S=200 D=OK F=sip:[REMOVED] T=sip:[EMAIL PROTECTED]
IP=10.2.252.181 [EMAIL PROTECTED]
New request - M=ACK RURI=sip:[EMAIL PROTECTED] F=sip:[REMOVED]
T=sip:[EMAIL PROTECTED] IP=10.2.252.190
[EMAIL PROTECTED]
Recording Route info
Loose route has returned true, attempting routing.
Setting up reply handler and relaying request
New request - M=BYE RURI=sip:[EMAIL PROTECTED]
F=sip:[EMAIL PROTECTED] T=sip:[EMAIL PROTECTED]
IP=10.2.252.190 [EMAIL PROTECTED]
Recording Route info
Loose route has returned true, attempting routing.
Setting up reply handler and relaying request
Reply - S=200 D=OK F=sip:[REMOVED] T=sip:[EMAIL PROTECTED]
IP=10.2.252.181 [EMAIL PROTECTED]
_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
