I didn't want to be sarcastic at all, just admired how you put your perspective that you fix a bug -- you are right and others are wrong. And yes, I am sure I am not aware of all SIP scenarios, every day I am surprised on what new issues I have to troubleshoot on SIP traffic. That's why I haven't listed all of them, just the case of BYE, but you actually prove my point of CANCEL and ACK -- if the servers active at CANCEL or ACK are not like for INVITE, is hard to ensure the same server is selected. A solution here could be to try to do some stickiness routing using htable.

I also agree that there could be different ways to achieve some new feature, but, imo, it is more clear to have a new algorithm that covers all your needs, instead of changing for all the other hashing algorithms, which are used by many others as they are. This brings more flexibility as well, one can use existing algorithms and the new one in the same config at the same time, not depending on a modparam.

Same approach was done when relative weight distribution algorithm was added, along with the exiting at that time, weight distribution algorithm.

You are more than welcome to propose PR to improve the documentation based on your findings and discussions on community forums/portals -- such contributions are very much appreciated!


You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or unsubscribe.