Feature Request: Unsafe Routes Active/Standby failover

[benyanke]

What version of nebula are you using? (nebula -version)

Latest

What operating system are you using?

Linux, but affects all

Describe the Bug

Currently if you define unsafe routes to provide a nebula network access to a non nebula subnet, you have introduced a single point of failure, the jump host (via).

I would propose adding a failover_via field with a list of IPs, which nebula would then use as a route if the primary via host is unavailable.

Note that this is similar to but distinct from #1332. 1332 allows HA in an active/active methodology when you have a BGP router available. This proposal would allow HA in an active/standby methodology with no router needed.

Logs from affected hosts

N/A

Config files from affected hosts

N/A

[Clement87]

Hi,
a quick implementation would be to use active/active methodology but setting one weight to 0.

Unfortunatelly we can't because of this control:

if gatewayWeight < 1 || gatewayWeight > math.MaxInt32 {
    return nil, fmt.Errorf("entry .weight in tun.unsafe_routes[%v].via[%v] is not in range (1-%d) : %v", i+1, ig+1, math.MaxInt32, gatewayWeight)
}

If you want to test the comportement, set a weight to 0 but instead of a restart, do a reload and the control is bypassed...
Configuration like this:

- route: 192.168.87.0/24
  via:
    - gateway: 10.0.0.1
      weight: 10
    - gateway: 10.0.0.2
      weight: 0

Then systemctl reload nebula