"HA" ie. two linked tincds on the same ethernet segment

hvjunk hvjunk at gmail.com
Wed Aug 17 00:32:35 CEST 2016


Goodday,

Currently using tinc 1.0.24 on Debian 8.x, OVH datacentre, two servers same VRACK (ie. internal ethernet bridge/switched together).

I’m wanting to get a remote server connected to both the two OVH servers for fail-over reasons, but having access via both to the shared vRack segment as a switch mode. Thereby Essentially I’m creating a ethernet loop…


----------------------vRack
  |                |
Server1   Server2
 |               |
 RemoteServer

tinc create tun/tap on server1 & Server2. this then connects to OpenVSwitch bridge/switch that have on both servers the eth1 interfaces as ports to that bridge/switch.


Questions:

1) Does tinc support some STP to pick up this type of loops and block one interface/link?
 I guess not, especially after looking at Broadcast = no|direct|mst

YEs, I might have a very corner case need which I am doing as a temporary solution while waiting for the right solution, but the question then gets raised:

2) Is there anybody that did any type of HA for tinc when the one daemon fails to connect to the other or vice versa?
(I guess my next stop will be something like the Linux HA if I really want this, else I’ll just have to manually switch between the two servers, or switch one off while the other is on…)


3) ANything else I might’ve missed this late in the evening?

Hendrik


More information about the tinc mailing list