[asterisk-bugs] [JIRA] Issue Comment Edited: (ASTERISK-20288) PhonerLite reports RTP read error when ICE Support Enabled

[Michael L. Young (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-20288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=196146#comment-196146 ] 

Michael L. Young edited comment on ASTERISK-20288 at 8/23/12 7:23 PM:
----------------------------------------------------------------------

Glad that I was thinking things through correctly.  That is why I stated that the error message was misleading and needed to be changed.  I saw the same thing that you just wrote about assuming errno was set.

I have been playing with Phonerlite and trying to figure out what is going on with it and Asterisk.  It would appear that Phonerlite doesn't support ICE (I don't see any settings or anything on the website about support for ICE).  So, Phonerlite doesn't send any ICE candidates when sending the INVITE to Asterisk.  I added some debug messages using functions from pjlib to find out what error message pj_ice_sess_on_rx_pkt is returning back and it is "Bad Request", which is correct since there was no ICE candidates in the SDP.

I have been reviewing rfc5245 and found section 5.1.  It would appear that we need to verify ICE support first before going through the ICE procedures.  Asterisk should parse the SDP on the initial request and if there are no ICE elements in the SDP, we should handle the call as if ICE is turned off for the peer.

      was (Author: elguero):
    Glad that I was thinking things through correctly.  That is why I stated that the error message was misleading and needed to be changed.  I saw the same thing that you just wrote about assuming errno was set.

I have been playing with Phonerlite and trying to figure out what is going on with it and Asterisk.  It would appear that Phonerlite doesn't support ICE (I don't see any settings or anything on the website about support for ICE).  So, Phonerlite doesn't send any ICE candidates when sending the INVITE to Asterisk.  Asterisk attempts to do ICE negotiation and when that fails, the call is hungup.  I added some debug messages using functions from pjlib to find out what error message pj_ice_sess_on_rx_pkt is returning back and it is "Bad Request", which is correct since there was no ICE candidates in the SDP.

I have been reviewing rfc5245 and found section 5.1.  It would appear that we need to verify ICE support first before going through the ICE procedures.  Asterisk should parse the SDP on the initial request and if there are no ICE elements in the SDP, we should handle the call as if ICE is turned off for the peer.
  
> PhonerLite reports RTP read error when ICE Support Enabled
> ----------------------------------------------------------
>
>                 Key: ASTERISK-20288
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-20288
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Resources/res_rtp_asterisk
>    Affects Versions: SVN, 11.0.0-beta1
>         Environment: Linux Ubuntu
>            Reporter: Joern Krebs
>            Severity: Minor
>         Attachments: debug, debug
>
>
> O.K. this is a bit hard to explain for me:
> If I use PhonerLite (and I can only replicate this issue with PhonerLite, but there it happens 100% of the time!) to call asterisk (any number, and any setup, nat, no-nat, etc.) I get the warning:
> RTP Read error: Success. Hanging up
> The "Success" part of the error should be the error message. BUT Success indicates, that there actually is no error.
> So what I did was changing the code and checking if the errno is 0, ignore that line ... and it works.
> I do not get this error with the SVN for Asterisk 10, just with SVN Asterisk 11.
> So basically under some circumstances (I cannot replicate the real issue, ... even the debug log doesn't show any issues at all!)
> the errno is set to 0, instead of whatever it should be...
> Can please someone have a look at this.
> (I would be open to test anything, although my dev capabilities are limited.)
> It really is a show-stopper to me! :-(

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira