Open robvanmieghem opened 5 years ago
started a full node with light client support and nat=none using that as bootnode however does not seem to work as it looks like it still uses the discovery from there or because the light client's enode url is the one from the nat . Will try the following: set nat to none in the light client, if that does not work, add the node directly as a peer. I must say that they are both running on the same machine so it might also recognize the connection as "self".
After testing b90bd76a1c1b09d6c22e2cd53847ad51a25744b7 At first glance, it looks like a bootnode does not propagate itself as a node to connect to
I had a full eth node and added it as bootnode with discovery enabled and without upnp. This seemed to work just fine. Unfortunately, pulled the disk from the server running this setup so resyncing everything. This config without a local bootnode seems not to sync automatically with the public ropsten network anymore. Best to add the possibility to override the ethbootnodes to tfchaind as well https://github.com/threefoldfoundation/tfchain/issues/283 (done)
Tested this setup again and works like a charm, this would mean we need to have a container in the network running a full eth node.
There are very few full nodes that support the experimental light client protocol as it is disabled by default. It looks like multiple connections from the same ip does not work.