Were there any changes to some regions recently that may have affected UDP traffic? For a while now, I’ve been using LiveKit in multiple regions. Sometime recently, some regions have stopped working.
It requires a broad UDP port range (50000-60000), but I’ve been using the trick that if you open one UDP port, all UDP traffic works. So my config just has:
[[services]]
http_checks = []
internal_port = 10000
protocol = "udp"
script_checks = []
[[services.ports]]
handlers = []
port = 10000
As of some time recently, I’m getting TCP traffic just fine, but my UDP connections are acting like they aren’t being routed.
Working: sea
, fra
Failing: iad
, ord
I know that I saw all of my ord
instances restart some time last night for multiple apps, so I’m wondering if there was a newer change to that and some other regions that has caused this to start failing. If so, is there a change I can make on my end to restore this or is there something that can be done to restore UDP ranges on Fly’s end?
For a test case, this is a guide for bringing up a LiveKit instance using a similar config: GitHub - bekriebel/livekit-flydotio: An example on how to deploy LiveKit on Fly.io