Manage peers
Hyperledger Besu peer-to-peer (P2P) discovery happens periodically based on the number of peers in a network and the node's peer limit.
The frequency of discovery isn't configurable, but you can limit remote connections in public networks and randomly prioritize connections in small, stable networks.
You can use admin_addPeer
to attempt a specific
connection, but this isn't P2P discovery.
In private networks, we recommend using bootnodes to initially discover peers.
P2P discovery process
The P2P discovery process requires ports to be open to UDP and TCP traffic. If you have a firewall in place, keep those ports open to allow traffic in and out. If you are running a node at home on your network, ensure that your router has those ports open.
The discovery
stack uses UDP to keep peer discovery lightweight and quick.
It only allows a node to find peers and connect to them, without any additional overhead.
Once peers have bonded, the data exchange between them is complex and needs a fully featured
protocol to support error checking and retries, so the devP2P
stack uses TCP.
Both stacks work in parallel: the discovery
stack adds new peers to the network, and the devP2P
stack enables interactions and data flow between them.
In detail, the P2P discovery process is as follows:
-
When Besu starts up it advertises its presence and details (including the enode) using UDP before establishing a formal connection with any peer (log messages look like
Enode URL enode://....
). -
Besu attempts to connect to the network's bootnodes (a set of predefined nodes used to help bootstrap discovery).
-
Once a connection with a bootnode is established using UDP (
ping/pong
handshake messages in the debug and trace logs), Besu requests a list of neighbors (potential peers) from the bootnode (find node
messages in the debug and trace logs). -
Besu attempts to connect to each peer using TCP, and get status information from them – such as network details, what the peer believes to be the current chain head, and its list of neighbors. From this point on any traffic to that peer is only done using TCP.
-
Depending on the synchronization method, a common block (the pivot block) is selected that all connected peers (default of 5) have, and Besu syncs from that block till it gets to chain head. Log messages look like
Downloading world state from peers for pivot block .......
. -
Besu repeats the same process for each peer in step 4, and any new peers that come along (regardless of client).
The more peers Besu is connected to, the more confident it is of having an accurate view of the network.
Limit peers
You can limit peers to reduce the bandwidth, CPU time, and disk access Besu uses to manage and respond to peers.
To reduce the maximum number of peers, use the --max-peers
option. The default is 25.
Limit remote connections
Prevent eclipse attacks when using --sync-mode
and --fast-sync-min-peers
on public networks by enabling the remote connection limits.
In private and permissioned networks with only trusted peers, enabling the remote connection limits is unnecessary and might adversely affect the speed at which nodes can join the network. Limiting remote connections can cause a closed group of peers to form when the number of nodes in the network is slightly higher than --max-peers
. The nodes in this closed group are all connected to each other and can't accept more connections.
You can use --random-peer-priority-enabled
to help prevent closed groups of peers in small, stable networks.
Monitor peer connections
JSON-RPC API methods to monitor peer connections include:
Each peer entry returned by admin_peers
includes a protocols
section. Use the information in the protocols
section to:
-
Determine the health of peers. For example, an external process can use
admin_peers
andadmin_removePeer
to disconnect from peers that are stalled at a single difficulty for an extended period of time. -
Monitor node health. For example, if peers report increasing difficulties but the node is stuck at the same block number, the node may be on a different fork to most peers.
-
Determine which protocol level peers are communicating with. For example, you can see if
"version": 65
is being used to reduce transaction sharing traffic.
List node connections
The default logging configuration doesn't list node connection and disconnection messages. To enable listing them, set the --logging
option to DEBUG
. For more verbosity, set the option to TRACE
.
The console logs connection and disconnection events when the log level is DEBUG
or higher. If the message Successfully accepted connection from ...
displays, connections are getting through the firewalls.
2018-10-16 12:37:35.479-04:00 | nioEventLoopGroup-3-1 | INFO | NettyP2PNetwork | Successfully accepted connection from 0xa979fb575495b8d6db44f750317d0f4622bf4c2aa3365d6af7c284339968eef29b69ad0dce72a4d8db5ebb4968de0e3bec910127f134779fbcb0cb6d3331163c
Disable discovery
To disable P2P discovery, set the --discovery-enabled
option to false
.
With discovery disabled, peers can't open connections with the node unless they were previously discovered or manually peered (for example, using admin_addPeer
). Static nodes can also open connections.
Troubleshoot
If you encounter issues with peering, see the troubleshoot peering documentation, which helps you identify and resolve common problems that can occur during the peering process.