Open AiyionPrime opened 3 years ago
There is no solution at the moment, except move coordinates a little bit. We nothing inside of leaflet that supports such a behavior.
Maybe you can add a plugin to your instance https://leafletjs.com/plugins.html#clusteringdecluttering
Probably it is easier to solve this on the respondd side. For our node map i solved it by introducing a pseudo Domain for all servers. I don't know if this is already possible with implementations like mesh-announce. We are using the not yet merged respondd implementation of yanic.
The yanic respondd config looks like this.
data_interval = "1m"
batman = [ "bat0", "bat1", ... ]
[[listen]]
address = "ff02::2:1001"
interface = "wg0"
[[listen]]
address = "ff02::2:1001"
interface = "vpn0"
port = 1001
[[listen]]
address = "ff02::2:1001"
interface = "vpn1"
port = 1001
....
[defaults]
node_id = ""
hostname = ""
site_code = "ffmwu"
domain_code = "servers"
vpn = true
interfaces_traffic = [ "eth0" ]
interfaces_address = [ "loopback" ]
The process listens on all mesh interfaces and on a direct Wireguard connection to the map server to make it mesh independent
As a result the gateways are shown only once on the map with all connected nodes.
Currently in our multidomainsetup supernodes exist 15 (domain amount) times in our setup. This looks a little weird in gravity-view but furthermore becomes unusable in normal map mode.
Especially when a supernode is emitting a location, all 15 instances of the same machine are stacked above aech other and make it a random game which domain is on top and clickable.
This is no easy task imo, as the current representation lacks either something like a merged view of all domains (https://github.com/ffnord/mesh-announce/issues/73), or an identifier of the machine, that's the same across its announced domains.
This could be based on @genofires efforts regarding the machine-id (https://github.com/ffnord/mesh-announce/issues/37), as this is primarily a supernode problem, which more than often come with systemd these days.
Is your feature request related to a problem?
The part, where supernodes are stacked on top of each other in the normal map is irritating.
Describe the solution you'd like
For now I only want to extend the discussion in hackint#gluon earlier this day.
Describe alternatives you've considered
An alternative could be to destack overlapping nodes on the normal map by something similar to the graph view, where nodes won't sit on top each other.
Additional context