blockbook

blockbook is currently in the state of heavy development, do not expect this documentation to be up to date

Build and install using docker

Run in the project root

make all

to create blockbook debian packages.

Install manually

Setup go environment (Debian 9):

sudo apt-get update && apt-get install -y \
    build-essential git wget pkg-config lxc-dev libzmq3-dev libgflags-dev libsnappy-dev zlib1g-dev libbz2-dev liblz4-dev
cd /opt
wget https://storage.googleapis.com/golang/go1.9.2.linux-amd64.tar.gz && tar xf go1.9.2.linux-amd64.tar.gz
sudo ln -s /opt/go/bin/go /usr/bin/go
go help gopath

Install RocksDB: https://github.com/facebook/rocksdb/blob/master/INSTALL.md and compile the static_lib and tools

git clone https://github.com/facebook/rocksdb.git
cd rocksdb
make release

Setup variables for gorocksdb: https://github.com/tecbot/gorocksdb

export CGO_CFLAGS="-I/path/to/rocksdb/include"
export CGO_LDFLAGS="-L/path/to/rocksdb -lrocksdb -lstdc++ -lm -lz -lbz2 -lsnappy -llz4"

Install ZeroMQ: https://github.com/zeromq/libzmq

Install go-dep tool:

RUN go get github.com/golang/dep/cmd/dep

Get blockbook sources, install dependencies, build:

cd $GOPATH/src
git clone https://github.com/trezor/blockbook.git
cd blockbook
dep ensure
go build

Usage

./blockbook --help

Example command

To run blockbook with fast synchronization, connection to ZeroMQ and providing https and socket.io interface, with database in local directory data and connected to local bitcoind at http://localhost:8332 with user rpc/rpc:

./blockbook -sync -parse -httpserver=127.0.0.1:8333 -socketio=127.0.01:8334 -certfile=server/testcert -zeromq=tcp://127.0.0.1:28332

Blockbook logs only to stderr, logging to files is disabled. Verbosity of logs can be tuned by command line parameters -v and -vmodule, details at https://godoc.org/github.com/golang/glog

Supported coins

• BTC
• BTC Testnet
• BCH
• BCH Testnet
• ZCash
• ZCash Testnet
• Ethereum
• Ethereum Ropsten Testnet

Data storage in RocksDB

Blockbook stores data the key-value store RocksDB. Data are stored in binary form to save space. The data are separated to different column families:

• default

  at the moment not used, will store statistical data etc.

• height - maps block height to block hash

  Block heigh stored as array of 4 bytes (big endian uint32) Block hash stored as array of 32 bytes

  Example - the first four blocks (all data hex encoded)

0x00000000 : 0x000000000933ea01ad0ee984209779baaec3ced90fa3f408719526f8d77f4943
0x00000001 : 0x00000000b873e79784647a6c82962c70d228557d24a747ea4d1b8bbe878e1206
0x00000002 : 0x000000006c02c8ea6e4ff69651f7fcde348fb9d557a06e6957b65552002a7820
0x00000003 : 0x000000008b896e272758da5297bcd98fdc6d97c9b765ecec401e286dc1fdbe10

• outputs - maps output script+block height to array of outpoints

  Output script (ScriptPubKey)+block height stored as variable length array of bytes for output script + 4 bytes (big endian uint32) block height array of outpoints stored as array of 32 bytes for transaction id + variable length outpoint index for each outpoint

  Example - (all data hex encoded)

0x001400efeb484a24a1c1240eafacef8566e734da429c000e2df6 : 0x1697966cbd76c75eb9fc736dfa3ba0bc045999bab1e8b10082bc0ba546b0178302
0xa9143e3d6abe282d92a28cb791697ba001d733cefdc7870012c4b1 : 0x7246e79f97b5f82e7f51e291d533964028ec90be0634af8a8ef7d5a903c7f6d301

• inputs - maps transaction outpoint to input transaction that spends it

  Transaction outpoint stored as array of 32 bytes for transaction id + variable length outpoint index Input transaction stored as array of 32 bytes for transaction id + variable length input index

  Example - (all data hex encoded)

0x7246e79f97b5f82e7f51e291d533964028ec90be0634af8a8ef7d5a903c7f6d300 : 0x0a7aa90ea0269c79f844c516805e4cac594adb8830e56fca894b66aab19136a428
0x7246e79f97b5f82e7f51e291d533964028ec90be0634af8a8ef7d5a903c7f6d301 : 0x4303a9fcfe6026b4d33ba488df6443c9a99bca7b7fcb7c6f6cd65cea24a749b700

Registry of ports

coin	blockbook http port	blockbook socket.io port	RPC port	zmq port
btc	9030	9130	8030	38330
bch	9031	9131	8031	38331
zec	9032	9132	8032	38332
dash	9033	9133	8033	38333
ltc	9034	9134	8034	38334
btg	9035	9135	8035	38335
eth	9036	9136	8036
etc	9037	9137	8037
xem	9038	9138	8038	38336
btc-testnet	19030	19130	18030	48330
bch-testnet	19031	19131	18031	48331
zec-testnet	19032	19132	18032	48332
eth-testnet	19036	19136	18036

Todo

• add db data version (column data version) checking to db to avoid data corruption
• improve txcache (time of storage, number/size of cached txs, purge cache)
• collect blockbook stats (number of items in indexes, etc)
• update documentation
• create/integrate blockchain explorer
• support all coins from https://github.com/trezor/trezor-common/blob/master/coins.json
• full ethereum support (tokens, balance)
• protobuf websocket interface instead of socket.io
• xpub index
• tests
• fix program dependencies to concrete versions
• protect socket.io interface against illicit usage
• optimize mempool (use non verbose get transaction, possibly parallelize)
• update used paths and users according to specification by system admin
• cleanup of the socket.io - do not send unnecessary data
• handle different versions of Bitcoin Core
• log live traffic from production bitcore server and replay it in blockbook
• find memory leak in initial import - disappeared with index v2
• zcash support
• basic ethereum support
• disconnect blocks - use block data if available to avoid full scan
• compute statistics of data, txcache, usage, etc.
• disconnect blocks - remove disconnected cached transactions
• implement getmempoolentry
• support altcoins, abstraction of blockchain server/service
• cache transactions in RocksDB
• parallel sync - rewrite - it is not possible to gracefully stop it now, can leave holes in the block
• mempool - return also input transactions
• blockchain - return inputs from mempool
• do not return duplicate txids
• legacy socket.io JSON interface
• disconnect blocks - optimize - full range scan is too slow and takes too much disk space (creates snapshot of the whole outputs), split to multiple iterators
• parallel sync - let rocksdb to compact itself from time to time, otherwise it consumes too much disk space