[OpenSIPS-Users] Opensips change the order of requests

Vlad Paiu vladpaiu at opensips.org
Fri Oct 23 10:57:23 CEST 2015


Hello,

I would say this an UA issue - OpenSIPS does not do any enforcing on the 
order of outgoing SIP packets, so if the UA sends an INVITE immediately 
followed by an UPDATE, there exists the risk of the UPDATE reaching the 
endpoint first ( even if the proxy would maintain the the packet order, 
since UDP is used as transport, there would still be no guarantee that 
the endpoint receives them in the same order ).

Can you configure the UA to only send the UPDATE after the INVITE 
transaction has finished ? This would be the only way to ensure there 
are no races.

Best Regards,

Vlad Paiu
OpenSIPS Developer

On 22.10.2015 20:05, Patrick Wakano wrote:
> Hello Opensips list!
>
> I already explained my scenario in this e-mail 
> (http://lists.opensips.org/pipermail/users/2015-October/032859.html) 
> where the Ack was propagated with wrong CSeq number when using the 
> in-dialog ping options. Vlad did a great job and already correct this 
> bug! (https://github.com/OpenSIPS/opensips/issues/680)
>
> Now in a scenario where I disabled the in-dialog ping, I faced another 
> issue. I mentioned that sometimes, due to unknown reasons, when I 
> received a Re-Invite immediately followed by an Update, the Update 
> gets forwarded first. The problem is that, since CSeq numbers are not 
> altered, when this inversion happens, my peer rejects the Re-Invite 
> because its CSeq number is lower than the Update one.
> Probably Opensips has to respect the ordering of in-dialog requests 
> and forward them accordingly, instead of handling them asynchronously 
> and forwarding right away...
>
> Best regards,
> Patrick
>
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20151023/eefd0e9e/attachment.htm>


More information about the Users mailing list