Node behind SonicWall Router

[Lance Fredrickson]

Recently the small business I work at replaced an aging linksys wrt 
style router with a SonicWall one. Ever since this change I have issues 
with this node falling back to TCP constantly. TCP for me ends up 
performing very sub-par, and most of the time unworkable. For some 
reason too the connection usually ends up being routed through another node.

I created a new tinc instance using TAP (instead of TUN) between this 
work PC, and one other host. I then added an additional route command in 
tinc-up so the work node is aware of the other tinc network. This seems 
to work much better with the new router. I'm guessing the constant flow 
and overhead (arps and such)  of TAP is helping to keep the sonicwall 
device from dropping/closing the connection.

I tried playing with PingInterval and lowered it to 30 seconds while 
previously using tun so the work node might stay better connected. I 
didn't see any better results, maybe even lower is necessary.  If I were 
to try to use tun again, would this be the route to go? lowering the 
ping time? It seems like the connection needs more activity for the 
connection not to be dropped.

I also have seen that running "tinc dump subnets" while using TAP 
outputs the MAC addresses tinc is aware. The MAC addresses leading zeros 
end up being truncated. Not really anything detrimental, but was just 
wondering if this was unintended, and thought I'd bring it up.

Thanks,

Lance