On 06/21/2012 10:34 PM, andrewpitman wrote:
Okay, I ran two versions of the Perl script against my test
server for a few hours last night, both of which loop and
send the crlf keepalive messages as fast as they could be
sent. One was sending UDP and the other TCP. I didn't
manage to get proxy to stop responding to registration
requests, but the CPU was pegged at about 125% on this dual
proc VM. I actually managed to fill up my /var filesystem
with all the junk going to sipXproxy.log, but didn't manage
to bring forth the issue...
Hi Andrew,
Yes you have a point. Modifying the for loop to something like below
should cover the situation you have pointed out
int crlfCount = 0;
for (int i = 0; i< res; i++)
{
if (readBuffer(i) == '\r' || readBuffer(i) == '\n')
{
crlfCount+++;
} else
{
break;
}
}
If you can generate your own rpms, feel free to modify and attach the patch to
the tracker. If you are not setup to build rpms, let me know, I'll compile a
new RPM withi this change. However, you will ahve to point your repo to stage
because there has been changes to some sipXtackLib header components recently
that would mean you need to update all c++ components that links to sipxtacklib.
_______________________________________________
sipx-users mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users/
