Problem in connecting new node to the network

Question

I am trying to setup a network of 2 nodes(node01, node02) and then want another node to be connected to that network.

Here is what my configuation is : For node 1:

[QUORUM_SET]
THRESHOLD_PERCENT=100
VALIDATORS=["$self", "$node02"]

For node 2:

[QUORUM_SET]
THRESHOLD_PERCENT=100
VALIDATORS=["$self", "$node01"]

I ran both of these nodes with --forcescp, they are connected and synced.

Now I want to connect my 3rd node with a condition : I trust on myself and either of the above 2 nodes. Here is the configuraiton which I did in node03:

# Trusting on my self
[QUORUM_SET]
THRESHOLD_PERCENT=100
VALIDATORS=["$self"]


# Trusting one of you two
[QUORUM_SET.1]
THRESHOLD_PERCENT=50
VALIDATORS=["$node01", "$node02"]

I ran the node03 with --focescp, but the node03 is being rejected and getting this error even if the handshake is successfull :

2019-06-12T18:30:30.307 GCOQO [Overlay INFO] successful handshake with [email protected]:11635
2019-06-12T18:30:30.322 GCOQO [Overlay WARNING] ignoring received localhost
2019-06-12T18:30:30.322 GCOQO [Overlay INFO] successful handshake with [email protected]:11625
2019-06-12T18:30:30.332 GCOQO [Overlay WARNING] Received error (ERR_LOAD): peer rejected
2019-06-12T18:30:30.333 GCOQO [Overlay INFO] Dropping peer [email protected]:11635
2019-06-12T18:30:30.333 GCOQO [Overlay WARNING] ignoring received localhost
2019-06-12T18:30:30.333 GCOQO [Overlay ERROR] TCPPeer::drop shutdown socket failed: Transport endpoint is not connected [TCPPeer.cpp:195]

what could possibly I am doing wrong here?

Answer 1

You must not forcescp any node except the first one to kickstart your network. Don't know for sure but I guess if you start an additional node with forcescp then it will pretty likely start validating a different ledger instead of synchronizing with the others and thus become incompatible to the other ones. Start your 3rd node from scratch (stellar-core new-db), make sure it is aware of your other peers (KNOWN_PEERS) and give it some time.

Besides for running an own network you need to also maintain your own history archive which needs to be accessable and configured on your 3rd node. A new (or long time offline) node catches up by utilizing the history.