[Home]

Summary:ASTERISK-12361: rtptimeout and rtpholdtimeout can only be set globally for users
Reporter:David Woolley (davidw)Labels:
Date Opened:2008-07-10 12:48:08Date Closed:2009-09-18 12:03:50
Priority:MinorRegression?No
Status:Closed/CompleteComponents:Channels/chan_sip/General
Versions:Frequency of
Occurrence
Related
Issues:
Environment:Attachments:
Description:rtptimeout and rtpholdtimeout can be set in the global SIP configuration, in which case they are honoured for incoming calls matched against a user entry.  They can also be specified for an individual device provided it is configured as a peer or friend.  In the case of a friend, they are not honoured when an incoming call is matched against the user part of the configuration.

This seems inconsistent, especially as it is meaningful and useful to apply such timeouts on calls that match as users.  In our case we want to ensure that an AgentLogin type agent gets logged out if they lose their connection, but we are wary of setting the timeouts globally, in case we have a connection that does silence suppression and doesn’t use RTCP.

Note that sip.conf.sample doesn’t list these parameters as allowed for users.  Nonetheless, I think that it would only make sense if were allowed.


****** ADDITIONAL INFORMATION ******

This can be demonstrated by pulling the network cable on the phone and looking for timeouts, however we also enhanced sip show channel to output the values of these timeouts and showed that they were being ignored, for an incoming call, when provided from the friend configuration, but set when provided globally.

For completeness, we also had the tentative patch for ASTERISK-12127, but this doesn’t work at anywhere near the RTP level.
Comments:By: Olle Johansson (oej) 2008-07-10 15:10:58

Note: Peers take incoming calls too and the RTP timers are valid in both directions of the media stream.

Anyway, this is a feature request. In trunk, this works already, since there's no more internal sip_user data structure.

By: Russell Bryant (russell) 2008-07-11 16:25:44

closing it out since this "feature" already exists in trunk