dcrdata

dcrdata

The dcrdata repository is a collection of golang packages and apps for Decred data collection, storage, and presentation.

Repository overview

../dcrdata              The dcrdata daemon.
├── blockdata           Package blockdata.
├── cmd
│   ├── rebuilddb       rebuilddb utility, for SQLite backend.
│   ├── rebuilddb2      rebuilddb2 utility, for PostgreSQL backend.
│   └── scanblocks      scanblocks utility.
├── dcrdataapi          Package dcrdataapi for golang API clients.
├── db
│   ├── dbtypes         Package dbtypes with common data types.
│   ├── dcrpg           Package dcrpg providing PostgreSQL backend.
│   └── dcrsqlite       Package dcrsqlite providing SQLite backend.
├── public              Public resources for block explorer (css, js, etc.).
├── explorer            Package explorer, powering the block explorer.
├── mempool             Package mempool.
├── rpcutils            Package rpcutils.
├── semver              Package semver.
├── stakedb             Package stakedb, for tracking tickets.
├── txhelpers           Package txhelpers.
└── views               HTML templates for block explorer.

Requirements

• Go 1.8.6 or newer.
• Running dcrd (>=1.1.2) synchronized to the current best block on the network.

Installation

Build from Source

The following instructions assume a Unix-like shell (e.g. bash).

• Install Go

• Verify Go installation:

  go env GOROOT GOPATH
  

• Ensure $GOPATH/bin is on your $PATH.

• Install dep, the dependency management tool.

  go get -u -v github.com/golang/dep/cmd/dep
  

• Clone the dcrdata repository. It must be cloned into the following directory.

  git clone https://github.com/dcrdata/dcrdata $GOPATH/src/github.com/dcrdata/dcrdata
  

• Fetch dependencies, and build the dcrdata executable.

  cd $GOPATH/src/github.com/dcrdata/dcrdata
  dep ensure
  # build dcrdata executable in workspace:
  go build
  

The sqlite driver uses cgo, which requires a C compiler (e.g. gcc) to compile the C sources. On Windows this is easily handled with MSYS2 (download and install MinGW-w64 gcc packages).

Tip: If you receive other build errors, it may be due to "vendor" directories left by dep builds of dependencies such as dcrwallet. You may safely delete vendor folders and run dep ensure again.

Runtime resources

Presently the dcrdata executable, it's config file, logs, data files, and web interface resources are all in the same folder. An option to specify the application data folder will be added in the future.

As with the config file, the "public" and "views" folders must be in the same folder as the dcrdata executable.

Updating

First, update the repository (assuming you have master checked out):

cd $GOPATH/src/github.com/dcrdata/dcrdata
git pull origin master
dep ensure
go build

Look carefully for errors with git pull, and reset locally modified files if necessary.

Getting Started

Create configuration file

Begin with the sample configuration file:

cp sample-dcrdata.conf dcrdata.conf

Then edit dcrdata.conf with your dcrd RPC settings. See the output of dcrdata --help for a list of all options and their default values.

Indexing the Blockchain

If dcrdata has not previously been run with the PostgreSQL database backend, it is necessary to perform a bulk import of blockchain data and generate table indexes. This will be done automatically by dcrdata, but the PostgreSQL tables may also be generated with the rebuilddb2 command line tool:

• Create the dcrdata user and database in PostgreSQL (tables will be created automatically).
• Set your PostgreSQL credentials and host in both ./cmd/rebuilddb2/rebuilddb2.conf and ./dcrdata.conf.
• Run rebuilddb2 -u to bulk import and index.
• In case of errors, or schema changes, the tables may be dropped with rebuilddb2 -D.

Starting dcrdata

Launch the dcrdata daemon and allow the databases to process new blocks. Both SQLite and PostgreSQL synchronization require about an hour the first time dcrdata is run, but they will be done concurrently. On subsequent launches, only blocks new to dcrdata are scanned.

./dcrdata    # don't forget to configure dcrdata.conf in this folder

The "public" and "views" folders must be in the same folder as the dcrdata executable.

dcrdata daemon

The root of the repository is the main package for the dcrdata app, which has several components including:

1. Block explorer (web interface).
2. Blockchain monitoring and data collection.
3. Mempool monitoring and reporting.
4. Data storage in durable database (sqlite presently).
5. RESTful JSON API over HTTP(S).

Block Explorer

After dcrdata syncs with the blockchain server via RPC, by default it will begin listening for HTTP connections on http://127.0.0.1:7777/. This means it starts a web server listening on IPv4 localhost, port 7777. Both the interface and port are configurable. The block explorer and the JSON API are both provided by the server on this port. See JSON REST API for details.

Note that while dcrdata can be started with HTTPS support, it is recommended to employ a reverse proxy such as nginx. See sample-nginx.conf for an example nginx configuration.

A new database backend using PostgreSQL was introduced in v0.9.0 that provides expanded functionality. However, initial population of the database takes additional time and tens of gigabytes of disk storage space. To disable the PostgreSQL backend (and the expanded functionality), dcrdata may be started with the --lite (-l for short) command line flag.

JSON REST API

The API serves JSON data over HTTP(S). All API endpoints are currently prefixed with /api (e.g. http://localhost:7777/api/stake), but this may be configurable in the future.

Endpoint List

The full ticket pool endpoints accept the URL query ?sort=[true\|false] for requesting the tickets array in lexicographical order. If a sorted list or list with deterministic order is not required, using sort=false will reduce server load and latency. However, be aware that the ticket order will be random, and will change each time the tickets are requested.

^*For the pool info block range endpoint that accepts the arrays url query, a value of true will put all pool values and pool sizes into separate arrays, rather than having a single array of pool info JSON objects. This may make parsing more efficient for the client.

All JSON endpoints accept the URL query indent=[true|false]. For example, /stake/diff?indent=true. By default, indentation is off. The characters to use for indentation may be specified with the indentjson string configuration option.

Important Note About Mempool

Although there is mempool data collection and serving, it is very important to keep in mind that the mempool in your node (dcrd) is not likely to be the same as other nodes' mempool. Also, your mempool is cleared out when you shutdown dcrd. So, if you have recently (e.g. after the start of the current ticket price window) started dcrd, your mempool will be missing transactions that other nodes have.

Command Line Utilities

rebuilddb

rebuilddb is a CLI app that performs a full blockchain scan that fills past block data into a SQLite database. This functionality is included in the startup of the dcrdata daemon, but may be called alone with rebuilddb.

rebuilddb2

rebuilddb2 is a CLI app used for maintenance of dcrdata's dcrpg database (a.k.a. DB v2) that uses PostgreSQL to store a nearly complete record of the Decred blockchain data. See the README.md for rebuilddb2 for important usage information.

scanblocks

scanblocks is a CLI app to scan the blockchain and save data into a JSON file. More details are in its own README. The repository also includes a shell script, jsonarray2csv.sh, to convert the result into a comma-separated value (CSV) file.

Helper packages

package dcrdataapi defines the data types, with json tags, used by the JSON API. This facilitates authoring of robust golang clients of the API.

package dbtypes defines the data types used by the DB backends to model the block, transaction, and related blockchain data structures. Functions for converting from standard Decred data types (e.g. wire.MsgBlock) are also provided.

package rpcutils includes helper functions for interacting with a rpcclient.Client.

package stakedb defines the StakeDatabase and ChainMonitor types for efficiently tracking live tickets, with the primary purpose of computing ticket pool value quickly. It uses the database.DB type from github.com/decred/dcrd/database with an ffldb storage backend from github.com/decred/dcrd/database/ffldb. It also makes use of the stake.Node type from github.com/decred/dcrd/blockchain/stake. The ChainMonitor type handles connecting new blocks and chain reorganization in response to notifications from dcrd.

package txhelpers includes helper functions for working with the common types dcrutil.Tx, dcrutil.Block, chainhash.Hash, and others.

Internal-use packages

Packages blockdata and dcrsqlite are currently designed only for internal use internal use by other dcrdata packages, but they may be of general value in the future.

blockdata defines:

• The chainMonitor type and its BlockConnectedHandler() method that handles block-connected notifications and triggers data collection and storage.
• The BlockData type and methods for converting to API types.
• The blockDataCollector type and its Collect() and CollectHash() methods that are called by the chain monitor when a new block is detected.
• The BlockDataSaver interface required by chainMonitor for storage of collected data.

dcrpg defines:

• The ChainDB type, which is the primary exported type from dcrpg, providing an interface for a PostgreSQL database.
• A large set of lower-level functions to perform a range of queries given a *sql.DB instance and various parameters.
• The internal package contains the raw SQL statements.

dcrsqlite defines:

• A sql.DB wrapper type (DB) with the necessary SQLite queries for storage and retrieval of block and stake data.
• The wiredDB type, intended to satisfy the APIDataSource interface used by the dcrdata app's API. The block header is not stored in the DB, so a RPC client is used by wiredDB to get it on demand. wiredDB also includes methods to resync the database file.

package mempool defines a mempoolMonitor type that can monitor a node's mempool using the OnTxAccepted notification handler to send newly received transaction hashes via a designated channel. Ticket purchases (SSTx) are triggers for mempool data collection, which is handled by the mempoolDataCollector class, and data storage, which is handled by any number of objects implementing the MempoolDataSaver interface.

Plans

See the GitHub issue tracker and the project milestones.

Contributing

Yes, please! See the CONTRIBUTING.md file for details, but here's the gist of it:

1. Fork the repo.
2. Create a branch for your work (git branch -b cool-stuff).
3. Code something great.
4. Commit and push to your repo.
5. Create a pull request.

Note that all dcrdata.org community and team members are expected to adhere to the code of conduct, described in the CODE_OF_CONDUCT file.

License

This project is licensed under the ISC License. See the LICENSE file for details.