Hello, do the self hosted “Global” discovery server can communicate with other servers if the node requested isn’t found ?
No, not currently. Syncthing should gain the ability to talk to more than one though, at some point.
How about changing the global discovery to a distributed hash table in the long term?
It’s been suggested a few times. I’m not convinced we gain that much by it, but if it’s a drop in replacement, why not.
Or swappable discovery plugins. Of course, you could do that for now by modifying the discovery server and running your own. Then you can have your server check multiple other servers or use DHT or Tor Hidden Services or smoke signals or whatever .
I do like the idea of a DHT so there isn’t a single point-of-failure (and just because), but I don’t think it’s a pressing issue right now. It’s easier to just allow for multiple discovery servers as backups. Anyone running Pulse in a mission-critical application should probably run their own discovery server anyway, or just hard-config the hosts.
Edit: I know DHT is not exactly “no single point-of-failure”, but it is somewhat more so. But again, I don’t think it’s a pressing issue.
I do believe the the OP wanted to ask if the discovery servers talk among themselves.
Client to Server A: Do you know YU87A…? Server A doesn’t Server A to Server B: Do you know YU87A… And so on…
Yes…[quote=“calmh, post:2, topic:1167, full:true”] No, not currently. Syncthing should gain the ability to talk to more than one though, at some point. [/quote]
…it was answered .
This idea might do it too. Thanks !