Some update here, to keep the list also informed.
I re-checked the code and I found some regression in CANCEL handling -
this was added last year with commit
4747da559f4df161441be8373488dee9fd16c282 when support for
"Content-Disposition: no-cancel" was added.
Ww - we have a record here :) - resuming a discussion over 8 years !
AFAIK, there was no intentional change (when comes to canceling branches
with no reply) - do you have a pcap + logs to show such behavior ?
And in regards to the sequence of CANCEL (on timeout) + forking, I think
I'm going to resurrect this as I've noticed that opensips 2.4.6 sends
CANCEL for branches it never got a provisional response for. Was this
changed intentionally?
Also, it looks like the discussion I am responding on was about opensips
1.5 but if any testing is needed around fail-over scenarios
Hi Piotr,
This sounds familiar to the problem I experienced some time ago - make
sure to check comments here:
https://sourceforge.net/tracker/?func=detailatid=1086410aid=2940556group_id=232389
I haven't been able to replicate that setup to confirm that the attached
patch works. You are
Hi guys,
Actually it will be great to have that patch tested to know for sure if
the problem is solved. I never got a 100% confirmation from Andrew, but
maybe Piotr can test and confirm.
Thanks and regards,
Bogdan
On 04/05/2011 04:58 PM, Andrew Pogrebennyk wrote:
Hi Piotr,
This sounds