Closed hradec closed 1 year ago
as a workaround, I have created a local.conf file where I set different custom secondary ports for each zerotier machine, and setup a crontab to open the custom ports in case they are not open, as it seems zerotier-one won't open custom ports automatically when they're setup on local.conf.
The patch above should fix both of your issues. Sounds like you've got a router with a funky UPnP implementation. A sane router should map a separate external port to 29994 for each machine like so:
The patch randomizes the port used for UPnP rather than making it deterministic and ending up with multiple on the same port. Also fixes the portmapper not running on manually specified tertiary ports.
What kind of router are these machines behind?
Closing this due to age and the suspicion that the issue was resolved with the patch above. Please re-open if this is still a problem. Thanks.
I have about 5 machines on a LAN network, all running zerotier (debian). I just noticed that port 29994 keeps been overwritten by each on those machines from time to time, which essentially means just one of those machines have direct UDP access from outside at a time.
It seems zerotier on linux doesn't check if port 29994 is already open to another ip, and just overwrites the redirection to itself all the time.
On the same lan I have 2 OSX machines running zerotier, and each of then choose a random UDP port to be open on the router, so they don't suffer of this problem.
What you expect to be happening.
What is actually happening?
Any steps to reproduce the error.
Any relevant console output or screenshots.
What operating system and ZeroTier version. Please try the latest ZeroTier release.