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.
├── api Package blockdata implements dcrdata's own HTTP API.
│ ├── insight Package insight implements the Insight API.
│ └── types Package types includes the exported structures used by
| the dcrdata and Insight APIs.
├── blockdata Package blockdata is the primary data collection and
| storage hub, and chain monitor.
├── cmd
│ ├── rebuilddb rebuilddb utility, for SQLite backend. Not required.
│ ├── rebuilddb2 rebuilddb2 utility, for PostgreSQL backend. Not required.
│ └── scanblocks scanblocks utility. Not required.
├── dcrdataapi Package dcrdataapi for golang API clients.
├── db
│ ├── agendadb Package agendadb is a basic PoS voting agenda database.
│ ├── dbtypes Package dbtypes with common data types.
│ ├── dcrpg Package dcrpg providing PostgreSQL backend.
│ └── dcrsqlite Package dcrsqlite providing SQLite backend.
├── dev Shell scripts for maintenance and deployment.
├── explorer Package explorer, powering the block explorer.
├── mempool Package mempool for monitoring mempool for transactions,
| data collection, and storage.
├── middleware Package middleware provides HTTP router middleware.
├── notification Package notification manages dcrd notifications, and
| synchronous data collection by a queue of collectors.
├── public Public resources for block explorer (css, js, etc.).
├── rpcutils Package rpcutils contains helper types and functions for
| interacting with a chain server via RPC.
├── semver Package semver.
├── stakedb Package stakedb, for tracking tickets.
├── testutil Package testutil provides some testing helper functions.
├── txhelpers Package txhelpers provides many functions and types for
| processing blocks, transactions, voting, etc.
├── version Package version describes the dcrdata version.
└── views HTML templates for block explorer.
Requirements
- Go 1.9.x or 1.10.x.
- Running
dcrd
(>=1.3.0) synchronized to the current best block on the
network. This is a strict requirement as testnet2 support is removed from
dcrdata v3.0.0.
- (Optional) PostgreSQL 9.6+, if running in "full" mode. v10.x is recommended
for improved dump/restore formats and utilities.
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. The current released binary of
dep
is recommended, but the latest
may be installed from git via:
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/decred/dcrdata $GOPATH/src/github.com/decred/dcrdata
-
Fetch dependencies, and build the dcrdata
executable.
cd $GOPATH/src/github.com/decred/dcrdata
dep ensure -vendor-only
# build dcrdata executable in workspace:
go build
To build with the git commit hash appended to the version, set it as follows:
go build -ldflags "-X github.com/decred/dcrdata/version.CommitHash=`git describe --abbrev=8 --long | awk -F "-" '{print $(NF-1)"-"$NF}'`"
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 -vendor-only
again.
Runtime resources
The config file, logs, and data files are stored in the application data folder,
which may be specified via the -A/--appdata
and -b/--datadir
settings.
However, the location of the config file may be set with -C/--configfile
. If
encountering errors involving file system paths, check the permissions on these
folders to ensure that the user running dcrdata is able to access these paths.
The "public" and "views" folders must be in the same folder as the dcrdata
executable. Set read-only permissions as appropriate.
Updating
First, update the repository (assuming you have master
checked out):
cd $GOPATH/src/github.com/decred/dcrdata
git pull origin master
dep ensure -vendor-only
go build
Look carefully for errors with git pull
, and reset locally modified files if
necessary.
Updating dependencies
dep ensure -vendor-only
fetches project dependencies, without making changes
in manifest files Gopkg.toml
and Gopkg.lock
.
Call dep ensure
to update project dependencies when introduce new imports.
For guides and reference materials about dep
, see the documentation.
The following FAQ explains the difference between Gopkg.toml
and Gopkg.lock
files.
Upgrading Instructions
Only necessary while upgrading from v2.x or below. The database scheme
change from dcrdata v2.x to v3.x does not permit an automatic migration. The
tables must be rebuilt from scratch:
- Drop the old dcrdata database, and create a new empty dcrdata database.
-- drop the old database
DROP DATABASE dcrdata;
-- create a new database with the same `pguser` set in the dcrdata.conf
CREATE DATABASE dcrdata OWNER dcrdata;
-- grant all permissions to user dcrdata
GRANT ALL PRIVILEGES ON DATABASE dcrdata to dcrdata;
-
Delete the dcrdata data folder (i.e. corresponding to the datadir
setting).
By default, datadir
is in {appdata}/data
:
- Linux:
~/.dcrdata/data
- Mac:
~/Library/Application Support/Dcrdata/data
- Windows:
C:\Users\<your-username>\AppData\Local\Dcrdata\data
(%localappdata%\Dcrdata\data
)
-
With dcrd synchronized to the network's best block, start dcrdata to begin
the initial block data import.
Getting Started
Configuring PostgreSQL (IMPORTANT)
If you intend to run dcrdata in "full" mode (i.e. with the --pg
switch), which
uses a PostgreSQL database backend, it is crucial that you configure your
PostgreSQL server for your hardware and the dcrdata workload.
Read postgresql-tuning.conf carefully for
details on how to make the necessary changes to your system. A helpful online
tool for determining good settings for your system is called
PGTune. DO NOT simply use this file in place
of your existing postgresql.conf or copy and paste these settings into the
existing postgresql.conf. It is necessary to edit postgresql.conf, reviewing all
the settings to ensure the same configuration parameters are not set in two
different places in the file.
On Linux, you may wish to use a unix domain socket instead of a TCP connection.
The path to the socket depends on the system, but it is commonly
/var/run/postgresql. Just set this path in pghost
.
Creating the Configuration File
Begin with the sample configuration file. With the default appdata
directory
for the current user on Linux:
cp sample-dcrdata.conf ~/.dcrdata/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
on a fresh startup.
Alternatively (but not recommended), 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
in the location specified by the appdata
flag.
- Run
./rebuilddb2
to bulk import data and index the tables.
- In case of irrecoverable errors, such as detected schema changes without an
upgrade path, the tables and their indexes may be dropped with
rebuilddb2 -D
.
Note that dcrdata requires that
dcrd is
running with some optional indexes enabled. By default, these indexes are not
turned on when dcrd is installed. To enable them, set the following in
dcrd.conf:
txindex=1
addrindex=1
If these parameters are not set, dcrdata will be unable to retrieve transaction
details and perform address searches, and will exit with an error mentioning
these indexes.
Starting dcrdata
Launch the dcrdata daemon and allow the databases to process new blocks. In
"lite" mode (without --pg
), only a SQLite DB is populated, which usually
requires 30-60 minutes. In "full" mode (with --pg
), concurrent synchronization
of both SQLite and PostgreSQL databases is performed, requiring from 3-12 hours.
See System Hardware Requirements for more
information.
On subsequent launches, only blocks new to dcrdata are processed.
./dcrdata # don't forget to configure dcrdata.conf in the appdata folder!
Unlike dcrdata.conf, which must be placed in the appdata
folder or explicitly
set with -C
, the "public" and "views" folders must be in the same folder as
the dcrdata
executable.
System Hardware Requirements
The time required to sync in "full" mode varies greatly with system hardware and
software configuration. The most important factor is the storage medium on the
database machine. An SSD (preferably NVMe, not SATA) is strongly recommended if
you value your time and system performance.
"lite" Mode (SQLite only)
Minimum:
- 1 CPU core
- 2 GB RAM
- HDD with 4GB free space
"full" Mode (SQLite and PostgreSQL)
These specifications assume dcrdata and postgres are running on the same machine.
Minimum:
- 1 CPU core
- 4 GB RAM
- HDD with 60GB free space
Recommend:
- 2+ CPU cores
- 7+ GB RAM
- SSD (NVMe preferred) with 60 GB free space
If PostgreSQL is running on a separate machine, the minimum "lite" mode
requirements may be applied to the dcrdata machine, while the recommended
"full" mode requirements should be applied to the PostgreSQL host.
dcrdata Daemon
The root of the repository is the main
package for the dcrdata
app, which
has several components including:
- Block explorer (web interface).
- Blockchain monitoring and data collection.
- Mempool monitoring and reporting.
- Database backend interfaces.
- RESTful JSON API (custom and Insight) 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 APIs 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 ("engine x"). See sample-nginx.conf for an
example Nginx configuration.
To save time and tens of gigabytes of disk storage space, dcrdata runs by
default in a reduced functionality ("lite") mode that does not require
PostgreSQL. To enable the PostgreSQL backend (and the expanded functionality),
dcrdata may be started with the --pg
switch.
Insight API (EXPERIMENTAL)
The Insight API is accessible via HTTP
at the /insight/api
URL path prefix, and via WebSocket at the
/insight/socket.io
URL path prefix.
The following endpoints are not implemented: status
, sync
, peer
, email
,
and rates
.
dcrdata API
dcrdata has its own JSON HTTP API in addition to the experimental Insight API
implementation. dcrdata's API endpoints are prefixed with /api
(e.g.
http://localhost:7777/api/stake
). The Insight API endpoints (not described in
this section) are prefixed with /insight/api
.
Endpoint List
Best block |
Path |
Type |
Summary |
/block/best |
types.BlockDataBasic |
Stake info |
/block/best/pos |
types.StakeInfoExtended |
Header |
/block/best/header |
dcrjson.GetBlockHeaderVerboseResult |
Hash |
/block/best/hash |
string |
Height |
/block/best/height |
int |
Size |
/block/best/size |
int32 |
Subsidy |
/block/best/subsidy |
types.BlockSubsidies |
Transactions |
/block/best/tx |
types.BlockTransactions |
Transactions Count |
/block/best/tx/count |
types.BlockTransactionCounts |
Verbose block result |
/block/best/verbose |
dcrjson.GetBlockVerboseResult |
Block X (block index) |
Path |
Type |
Summary |
/block/X |
types.BlockDataBasic |
Stake info |
/block/X/pos |
types.StakeInfoExtended |
Header |
/block/X/header |
dcrjson.GetBlockHeaderVerboseResult |
Hash |
/block/X/hash |
string |
Size |
/block/X/size |
int32 |
Subsidy |
/block/best/subsidy |
types.BlockSubsidies |
Transactions |
/block/X/tx |
types.BlockTransactions |
Transactions Count |
/block/X/tx/count |
types.BlockTransactionCounts |
Verbose block result |
/block/X/verbose |
dcrjson.GetBlockVerboseResult |
Block H (block hash) |
Path |
Type |
Summary |
/block/hash/H |
types.BlockDataBasic |
Stake info |
/block/hash/H/pos |
types.StakeInfoExtended |
Header |
/block/hash/H/header |
dcrjson.GetBlockHeaderVerboseResult |
Height |
/block/hash/H/height |
int |
Size |
/block/hash/H/size |
int32 |
Subsidy |
/block/best/subsidy |
types.BlockSubsidies |
Transactions |
/block/hash/H/tx |
types.BlockTransactions |
Transactions count |
/block/hash/H/tx/count |
types.BlockTransactionCounts |
Verbose block result |
/block/hash/H/verbose |
dcrjson.GetBlockVerboseResult |
Block range (X < Y) |
Path |
Type |
Summary array for blocks on [X,Y] |
/block/range/X/Y |
[]types.BlockDataBasic |
Summary array with block index step S |
/block/range/X/Y/S |
[]types.BlockDataBasic |
Size (bytes) array |
/block/range/X/Y/size |
[]int32 |
Size array with step S |
/block/range/X/Y/S/size |
[]int32 |
Transaction T (transaction id) |
Path |
Type |
Transaction details |
/tx/T |
types.Tx |
Transaction details w/o block info |
/tx/trimmed/T |
types.TrimmedTx |
Inputs |
/tx/T/in |
[]types.TxIn |
Details for input at index X |
/tx/T/in/X |
types.TxIn |
Outputs |
/tx/T/out |
[]types.TxOut |
Details for output at index X |
/tx/T/out/X |
types.TxOut |
Vote info (ssgen transactions only) |
/tx/T/vinfo |
types.VoteInfo |
Serialized bytes of the transaction |
/tx/hex/T |
string |
Same as /tx/trimmed/T |
/tx/decoded/T |
types.TrimmedTx |
Transactions (batch) |
Path |
Type |
Transaction details (POST body is JSON of types.Txns ) |
/txs |
[]types.Tx |
Transaction details w/o block info |
/txs/trimmed |
[]types.TrimmedTx |
Address A |
Path |
Type |
Summary of last 10 transactions |
/address/A |
types.Address |
Number and value of spent and unspent outputs |
/address/A/totals |
types.AddressTotals |
Verbose transaction result for last 10 transactions |
/address/A/raw |
types.AddressTxRaw |
Summary of last N transactions |
/address/A/count/N |
types.Address |
Verbose transaction result for last N transactions |
/address/A/count/N/raw |
types.AddressTxRaw |
Summary of last N transactions, skipping M |
/address/A/count/N/skip/M |
types.Address |
Verbose transaction result for last N transactions, skipping M |
/address/A/count/N/skip/Mraw |
types.AddressTxRaw |
Stake Difficulty (Ticket Price) |
Path |
Type |
Current sdiff and estimates |
/stake/diff |
types.StakeDiff |
Sdiff for block X |
/stake/diff/b/X |
[]float64 |
Sdiff for block range [X,Y] (X <= Y) |
/stake/diff/r/X/Y |
[]float64 |
Current sdiff separately |
/stake/diff/current |
dcrjson.GetStakeDifficultyResult |
Estimates separately |
/stake/diff/estimates |
dcrjson.EstimateStakeDiffResult |
Ticket Pool |
Path |
Type |
Current pool info (size, total value, and average price) |
/stake/pool |
types.TicketPoolInfo |
Current ticket pool, in a JSON object with a "tickets" key holding an array of ticket hashes |
/stake/pool/full |
[]string |
Pool info for block X |
/stake/pool/b/X |
types.TicketPoolInfo |
Full ticket pool at block height or hash H |
/stake/pool/b/H/full |
[]string |
Pool info for block range [X,Y] (X <= Y) |
/stake/pool/r/X/Y?arrays=[true|false] * |
[]apitypes.TicketPoolInfo |
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.
Vote and Agenda Info |
Path |
Type |
The current agenda and its status |
/stake/vote/info |
dcrjson.GetVoteInfoResult |
Mempool |
Path |
Type |
Ticket fee rate summary |
/mempool/sstx |
apitypes.MempoolTicketFeeInfo |
Ticket fee rate list (all) |
/mempool/sstx/fees |
apitypes.MempoolTicketFees |
Ticket fee rate list (N highest) |
/mempool/sstx/fees/N |
apitypes.MempoolTicketFees |
Detailed ticket list (fee, hash, size, age, etc.) |
/mempool/sstx/details |
apitypes.MempoolTicketDetails |
Detailed ticket list (N highest fee rates) |
/mempool/sstx/details/N |
apitypes.MempoolTicketDetails |
Other |
Path |
Type |
Status |
/status |
types.Status |
Coin Supply |
/supply |
types.CoinSupply |
Endpoint list (always indented) |
/list |
[]string |
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 exactly
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. This functionality is included in the startup of the
dcrdata daemon, but may be called alone with rebuilddb. 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 DataSourceLite
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:
- Fork the repo.
- Create a branch for your work (
git branch -b cool-stuff
).
- Code something great.
- Commit and push to your repo.
- Create a pull request.
DO NOT merge from master to your feature branch; rebase.
Before committing any changes to the Gopkg.lock file, you must update dep
to
the latest version via:
go get -u github.com/golang/dep/cmd/dep
To update dep
from the network, it is important to use the -u
flag as
shown above.
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.
Also, come chat with us on Slack!
License
This project is licensed under the ISC License. See the LICENSE file for details.