|Summary:||ASTERISK-20008: outboundproxy ignored after when sending invite after 407|
|Reporter:||Marcus Hunger (fnordian)||Labels:|
|Date Opened:||2012-06-15 11:26:35||Date Closed:||2012-06-19 10:39:20|
|Versions:||22.214.171.124 10.3.1||Frequency of|
|Environment:||Attachments:||( 0) ASTERISK-20008.patch|
( 1) invite.pcap
|Description:||When placing a call to a sip-peer which has an outboundproxy configured, the first invite is sent to the outboundproxy but the second one, after 407 is sent to the host specified in the request uri. This behavior was introduced in r356476 and is wrong in my opinion.|
|Comments:||By: Matt Jordan (mjordan) 2012-06-15 11:35:59.196-0500|
What were the versions of Asterisk you tested this with?
Can you provide a pcap of the message traffic?
By: Marcus Hunger (fnordian) 2012-06-15 11:50:50.711-0500
I tested with 10.3. To narrow it down to the specific revision I also tested 10.5.1-r356476. Same setup works without error on 10.5.1-r356475.
By: Marcus Hunger (fnordian) 2012-06-15 11:52:05.265-0500
pcap showing that the second invite is sent to the wrong host.
By: Matt Jordan (mjordan) 2012-06-15 11:56:23.468-0500
I'm confused: when it worked, did you test with 10.5.1 without r356476? Or did you test with a check-out of the 10 branch, prior to that revision?
By: Marcus Hunger (fnordian) 2012-06-15 12:01:28.410-0500
r356476 introduced the bug. just to confirm that, i tried the previous revision (r356475). it worked fine with the old one. it did not (invite was sent wrong) with the new one.
By: Mark Michelson (mmichelson) 2012-06-15 14:04:26.964-0500
In your pcap, it looks like the ACK is routed incorrectly too. I think the problem is that the destination for the INVITE needs to stay the same until a 200 OK is finally received. The patch should result in both the ACK and the ensuing INVITE being sent to the outbound proxy.
By: Marcus Hunger (fnordian) 2012-06-18 02:51:55.619-0500
I've tried that change already and it worked for me.
By: Walter Doekes (wdoekes) 2012-06-19 10:09:18.088-0500
/me points to ASTERISK-19677
and to https://reviewboard.asterisk.org/r/1859/diff/#0.33