Hi Team,
I am experiencing call lag during warm transfers after the agent leaves the room and only the human participants remain on the call. The lag affects the entire conversation, causing delays and sometimes resulting in participants talking over each other.
We are using LiveKit Cloud, and our SIP server is hosted in Australia.
Could anyone help identify the possible causes of this issue?
My first instinct would be that your users are in Australia and your room is elsewhere, causing the lag / delays. If not that, then something in your media pipeline is causing a very long hop.
I recommend taking a look through this doc which explains the various aspects of a regional deployment:
If you share a session ID or IDs (starts with RM_) then I can also take a look to see if I can see anything.
Thanks! Yes — our callers are all based in Australia, so a cross-region media hop is very plausible. I read through the regional deployment checklist.
Here’s a call ID for one affected call: SCL_FQzBnjPHbuHW
A quick question:
Our traffic is inbound SIP — what’s the right way to keep that in-region for AU callers?
The media for that call is passing through our Mumbai region. I think this is happening because the SIP trunk, ST_tLutTKCFtHXg placed the call on the Mumbai node. The first thing to check would be to region pin that outbound trunk, Region pinning for telephony | LiveKit Documentation , then I would expect the media to then route through Australia.
Our traffic is inbound SIP — what’s the right way to keep that in-region for AU callers?
The call ID you gave me looks like an outbound call (originated from LiveKit). For inbound Sip, please scroll up on that link above and it describes using the regional endpoints
Also, since you are on Scale, you could guarantee that all your media stays within AUS by pinning your LiveKit server