tinc - Jan 2019 - debugging, dns resolving not stable with tinc to linux bridge and increasing clients

Hello everybody,

I am starting to debug a issue I have with a tinc server.

Tinc server is connected to a Linux bridge br0 to a local network and 
every Tinc client connects with DHCP over this bridge to the local network.

DNS and everything else are going over the tinc vpn connection.

The past few months the clients (all windows) are reporting that DNS 
resolving stops working and they have to restart the TINC client and the 
TINC server to make it work again. The number of clients and traffic has 
increased the past few months and I am wondering what bottle neck may be 
causing this behaviour, the clients are connected and they have access 
to the Windows domain and file server, but DNS resolving stops working.

# brctl showmacs br0 | wc -l
116

Their are no overflow messages from the kernel.

The only errors I can find are "Metadata socket read errors"

net.ipv4.neigh.default.gc_thresh1 = 128
net.ipv4.neigh.default.gc_thresh2 = 512
net.ipv4.neigh.default.gc_thresh3 = 1024
net.ipv4.neigh.default.gc_interval = 30
net.ipv4.neigh.default.gc_stale_time = 60

This is a ipv4 only TINC server.

Any ideas?

Kind regards and best wishes,

Jelle de Jong (GNU/Linux Consultant)
PowerCraft Technology
I: www.powercraft.nl
T: +3185 060 9913
M: +316 1218 2441