Repository navigation
lighthouse high-availability with redundancy #1934
|
it seems we can configure HA for a lighthouse in two different ways
I am wondering if the second setup (2) is correct or no? I have tested both , both working, Regards |
Replies: 1 comment
|
Hello @shakibamoshiri - Use option 1. You should not reuse Nebula (overlay) IPs in your network, including for lighthouses that don't have their own overlay traffic (e.g. Nebula tracks peers (including lighthouses) by their overlay IP address. Nebula keeps a single tunnel per overlay IP at a time. This is visible in the static_host_map where peers are keyed by their Nebula IP. If you list multiple remote addresses that point to different machines under one IP, Nebula will only be connected to one of them at a time. Lighthouses do not share state with each other. In option 1, each host reports its addresses to every lighthouse in More generally, any two hosts that need to communicate must share at least one live lighthouse. They don't need to connect to every lighthouse. Picture a Venn diagram where every host uses 2 of 3 lighthouses, so if one goes down, each pair of hosts still has at least one lighthouse in common. |
Hello @shakibamoshiri -
Use option 1. You should not reuse Nebula (overlay) IPs in your network, including for lighthouses that don't have their own overlay traffic (e.g.
tun.disabled: true).Nebula tracks peers (including lighthouses) by their overlay IP address. Nebula keeps a single tunnel per overlay IP at a time. This is visible in the static_host_map where peers are keyed by their Nebula IP. If you list multiple remote addresses that point to different machines under one IP, Nebula will only be connected to one of them at a time.
Lighthouses do not share state with each other. In option 1, each host reports its addresses to every lighthouse in
lighthouse.hostsand queries all of the…