| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
ssh://shub-niggurath.zerotier.com:222/git/ZeroTierOne into adamierymenko-dev
|
|
|
|
test.zerotier.com net instead of the live net.
|
|
of the old size prefix way.
|
|
|
|
based on recommendations.
|
|
mi-go.zerotier.com (Singapore) due to real-world usage data.
|
|
|
|
|
|
since doing so can result in asymmetric failed NAT-t over UDP if one side has a firewall that permits outgoing UDP but not incoming.
|
|
|
|
UDP preference in send().
|
|
VERB_RENDEZVOUS computation.
|
|
longer deleting taps on Windows!
|
|
|
|
|
|
|
|
|
|
|
|
|
|
tap driver. Seems to work. Now to see if the cleanup we did here gets rid of the zombie tap device issue on Windows.
|
|
what we did this for: a refactor of the Windows tap connector.
|
|
|
|
pretty well now!
|
|
|
|
|
|
non-privileged users can access them (if they have auth token).
|
|
authtoken.secret if system unreadable.
|
|
|
|
supernodes from resynchronizing unless explicitly ordered.
|
|
and there is no point in mallocing the TCP write buffer.
|
|
|
|
|
|
|
|
indeed learn incoming TCP paths. Otherwise the recipient of a TCP connection does not know to reply via TCP! Heh.
|
|
certificates of membership tagged onto them. This likely will not actually increase the number of fragments in practice much, since these are then subsequently compressed. But the buffer needs to be allocated with room for them, otherwise these packets get dropped with a range_error exception internally.
|
|
|
|
|
|
|
|
|
|
done on Windows (I think).
|
|
|
|
|
|
|
|
|
|
|
