The Nexa Peer-to-Peer (P2P) Network protocol is a binary protocol used by Full Nodes and SPV Nodes, transmitted over TCP/IP.
Individual nodes on the Nexa network connect and create a mesh network where each node is indirectly connected to many others via just a couple of hops.
In the original Satoshi implementation of the P2P protocol the design of INV and getdata have been used for propagating transaction data using the rules of the gossip protocol values: forwarding validated transactions to a few peer-nodes who send it to others until the entire network has the transaction.
This emergent behavior of the P2P layer allows fast propagation without undue strain on any individual node.
The P2P protocol is designed around messages.
Each message is separate and self-contained.
Nodes should be tolerant of message-types they do not understand.
It is best to simply ignore those.
Generally speaking, each message is an event that the node can choose to respond to.
Events can be notifications of new data (transactions/blocks/etc), requests for such data to be sent, or the sending of the data itself.
In some specific cases a message can indicate the rejection of another message, though this is optional and should not be relied upon.
These design decisions were made with consideration to communication with untrusted/uncooperative partners.
Developer Notes: A common message strategy is to wait for any message that provides the required data (with a timeout), and then separately issue the request in a retry loop to multiple peers.
The P2P network has a variety of message types.
All P2P messages follow a binary format with the following structure:
|net magic||4 bytes||byte array(BE)||See net magic.|
|command string||12 bytes||string(BE)||See command string.|
|payload byte count||4 bytes||unsigned integer(LE)||The size of the payload. The total max size of any message is
|payload checksum||4 bytes||byte array(BE)||The message checksum is the first 4 bytes of a double-sha256 hash of the payload.|
|payload||variable||message-specific||See message types for links to message-specific page, which describe the payload for each message.|
See Example Message for a concrete example of this with a message that does not contain an extended payload.
The network identifier is used to separate blockchains and test networks.
This reduces unnecessary load on peers, allowing them to rapidly ban nodes rather then forcing the peer to do a blockchain analysis before banning or disconnecting.
For Nexa main net, the
net magic field is always
Any message received that does not begin with the
net magic is invalid.
net magic is designed to be unlikely to occur in normal data–the characters are rarely used upper ASCII, are not valid as UTF-8, and produce a large 32-bit integer with any alignment.
command string is a fixed-length 12 byte ASCII string.
Commands may not be longer than 12 bytes.
Commands that are shorter than 12 bytes are right-padded with null bytes (
The command string is used to determine the type of message being transmitted.
Messages with an unrecognized
command string are ignored by most implementations but may result in a ban by implementations that diverge from the Satoshi-client defacto standard.
|filteradd||Adds a single item into an installed filter|
|filterclear||Removes an installed filter|
|filterload||Inserts a transaction and merkle block filter into the peer|
|inv||Notifies peers about the existence of some information (generally a block or transaction)|
|dsproof-beta||Informs participants of an attempt to double spend|
|feefilter||Requests that transactions without sufficient fees are not relayed|
|getaddr||Requests a list of active peers|
|getblocks||Requests block hash identifiers|
|getdata||Requests information from a peer|
|getheaders||Requests block headers from a peer|
|ping||Requests a confirmation (pong) that the peer is still active|
|sendheaders||Requests that new blocks are sent as headers instead of hashes|
|version||Describes peer capabilities, particularly through the Services Bitfield|
|xversion||Describes peer capabilities in an extensible manner.|
|xupdate||Communicates a change in peer capabilities.|
|mempool||Request mempool contents|
|addr||Provides a peer with the addresses of other peers|
|block||Provides the contents of a block|
|headers||Provides a set of block headers (unsolicited or GETHEADERS response)|
|notfound||Indicates that a requested resource could not be relayed|
|merkleblock||Provides a provable subset of a block’s transactions, as filtered by FILTERADD|
|pong||Reply to a ping message|
|reject||Response by well-behaved clients if a message cannot be handled|
|tx||Provides a transaction|
|verack||Response to a version message|
Compact blocks, defined in BIP-152, seek to minimize the amount of data transferred when a block is mined by taking advantage of the fact that peers often already have most, if not all, of the transactions in a new block.
|sendcmpct||Indicates that this node supports the Compact Block protocol.|
|cmpctblock||Announces and provides abbreviated contents of a block.|
|getblocktxn||Requests additional transactions from a given block.|
|blocktxn||Returns requests transactions contained within a block (in response to getblocktxn.|
|get_xblocktx||Request unknown transactions from a block.|
|get_xthin||Request a previously announced xthin block from the announcing peer.|
|thinblock||A description of a block including full transactions only when it is known that the peer does not have them.|
|xthinblock||A description of a block including full transactions only when it is known that the peer does not have them. Uses truncated hashes to minimize data transfer|
|xblocktx||Provides a set of a transactions contained within a block (in response to get_xblocktx).|
The below segments, when concatenated in order, create a sample verack message.
|Label||Sample Value (Hexadecimal Representation)|
|Command String (“verack”)(BE)||
|Payload Byte Count(LE)||
Below is the full, concatenated sample message (in hexadecimal):