Hi, I know I can work around this issue, but was wondering if the behaviour is expected… If I use an application that blocks, mod_http_cache for example and the web server is unavailable - the call will stay up until the connect-timeout is reached. Obviously I can set a lower connect-timeout, but this doesn’t really resolve the issue. So, say I set connect-timeout to 3 seconds and place a 1 second call via soft phone (caller hangup), the billable duration will be 3 seconds - should the billable duration not be 1 second to match the calling party? also, shouldn’t a hangup force any running applications to stop? Again, obviously I can avoid this by being a bit smarter with the usage of http_cache, etc - but was really wondering if a) thats expected behaviour and the callee should count the duration based on the ACK from the BYE (which I suspect non do), or b) maybe there is a duration variable that I’ve not yet seen that contains the call duration until we receive the BYE from caller?. I would have thought (and Im probably wrong), billable duration s/u/m should resemble the SIP durations, i.e. endtime = time BYE was received? The other byproduct of this symptom is that the channels will stack up until call-timeout is reached - if it was left to the default of 300 seconds - a dead web server could soon cause the switch to burn out. Again, easily resolvable by not using blocking applications, doing things in the background, etc but… Thoughts & postcards? Thanks Oz.