A Node.js driver for RethinkDB with more advanced features.
npm install rethinkdbdash
Note: The rethinkdbdash-unstable
package is a relic from the past (rethinkdb < 1.13).
Rethinkdbdash uses almost the same API as the official driver. Please refer to the official driver's documentation for all the ReQL methods (the methods used to build the query).
The main differences are:
- You need to execute the module when you import it:
var r = require('rethinkdbdash')();
// With the official driver:
// var r = require('rethinkdb');
- Connections are managed by the driver with an efficient connection pool.
Once you have imported the driver, you can immediately run queries,
you don't need to call
r.connect
, or pass a connection torun
.
var r = require('rethinkdbdash')();
r.table('users').get('orphee@gmail.com').run().then(function(user) {
// ...
}).error(handleError)
- Cursors are coerced to arrays by default
var r = require('rethinkdbdash')();
r.table('data').run().then(function(result) {
assert(Array.isArray(result)) // true
// With the official driver you need to call
// result.toArray().then(function(result2) {
// assert(Array.isArray(result2))
// })
});
You can replace the official driver with rethinkdbdash by just replacing
var r = require('rethinkdb');
With:
var r = require('rethinkdbdash')({
pool: false,
cursor: true
});
If you want to take advantage of the connection pool, refer to the next section.
To switch from the official driver to rethinkdbdash and get the most of it, here are the few things to do:
- Change the way to import the driver.
var r = require('rethinkdb');
To:
var r = require('rethinkdbdash')();
// Or if you do not connect to the default local instance:
// var r = require('rethinkdbdash')({servers: [{host: ..., port: ...}]});
- Remove everything related to a connection:
r.connect({host: ..., port: ...}).then(function(connection) {
connection.on('error', handleError);
query.run(connection).then(function(result) {
// console.log(result);
connection.close();
});
});
Becomes:
query.run().then(function(result) {
// console.log(result);
});
- Remove the methods related to the cursor. This typically involves
removing
toArray
:
r.table('data').run(connection).then(function(cursor) {
cursor.toArray().then(function(result) {
// console.log(result):
});
});
Becomes
r.table('data').run().then(function(result) {
// console.log(result);
});
Note: Support for a TLS proxy is experimental.
RethinkDB does not support TLS connections to the server yet, but in case you want to run it over an untrusted network and need encryption, you can easily run a TLS proxy on your server with:
var tls = require('tls');
var net = require('net');
var tlsOpts = {
key: '', // You private key
cert: '' // Public certificate
};
tls.createServer(tlsOpts, function (encryptedConnection) {
var rethinkdbConn = net.connect({
host: 'localhost',
port: 28015
});
encryptedConnection.pipe(rethinkdbConn).pipe(encryptedConnection);
}).listen(29015);
And then safely connect to it with the tls
option:
var r = require('rethinkdbdash')({
port: 29015,
host: 'place-with-no-firewall.com',
ssl: true
});
ssl
may also be an object that will be passed as the options
argument to
tls.connect
.
Rethinkdbdash ships with a few interesting features.
When you import the driver, as soon as you execute the module, you will create
a default connection pool (except if you pass {pool: false}
. The options you
can pass are:
db
:<string>
- The default database to use if none is mentioned.user
:<string>
- The RethinkDB user, default value is admin.password
:<string>
- The password for the user, default value is an empty string.discovery
:<boolean>
- When true, the driver will regularly pull data from the tableserver_status
to keep a list of updated hosts, defaultfalse
pool
:<boolean>
- Set it tofalse
, if you do not want to use a connection pool.buffer
:<number>
- Minimum number of connections available in the pool, default50
max
:<number>
- Maximum number of connections available in the pool, default1000
timeout
:<number>
- The number of seconds for a connection to be opened, default20
pingInterval
: - if> 0
, the connection will be pinged everypingInterval
seconds, default-1
timeoutError
:<number>
- Wait time before reconnecting in case of an error (in ms), default 1000timeoutGb
:<number>
- How long the pool keep a connection that hasn't been used (in ms), default 60*60*1000maxExponent
:<number>
- The maximum timeout before trying to reconnect is 2^maxExponent x timeoutError, default 6 (~60 seconds for the longest wait)silent
: - console.error errors, defaultfalse
servers
: an array of objects{host: <string>, port: <number>}
representing RethinkDB nodes to connect tooptionalRun
: - iffalse
, yielding a query will not run it, defaulttrue
log
: - will be called with the log events by the pool master
In case of a single instance, you can directly pass host
and port
in the top level parameters.
Examples:
// connect to localhost:8080, and let the driver find other instances
var r = require('rethinkdbdash')({
discovery: true
});
// connect to and only to localhost:8080
var r = require('rethinkdbdash')();
// Do not create a connection pool
var r = require('rethinkdbdash')({pool: false});
// Connect to a cluster seeding from `192.168.0.100`, `192.168.0.101`, `192.168.0.102`
var r = require('rethinkdbdash')({
servers: [
{host: '192.168.0.100', port: 28015},
{host: '192.168.0.101', port: 28015},
{host: '192.168.0.102', port: 28015},
]
});
// Connect to a cluster containing `192.168.0.100`, `192.168.0.100`, `192.168.0.102` and
// use a maximum of 3000 connections and try to keep 300 connections available at all time.
var r = require('rethinkdbdash')({
servers: [
{host: '192.168.0.100', port: 28015},
{host: '192.168.0.101', port: 28015},
{host: '192.168.0.102', port: 28015},
],
buffer: 300,
max: 3000
});
You can also pass {cursor: true}
if you want to retrieve RethinkDB streams as cursors
and not arrays by default.
Note: The option {stream: true}
that asynchronously returns a stream is deprecated. Use toStream
instead.
Note: The option {optionalRun: false}
will disable the optional run for all instances of the driver.
Note: Connections are created with TCP keep alive turned on, but some routers seem to ignore this setting. To make
sure that your connections are kept alive, set the pingInterval
to the interval in seconds you want the
driver to ping the connection.
Note: The error __rethinkdbdash_ping__
is used for internal purposes (ping). Do not use it.
As mentioned before, rethinkdbdash
has a connection pool and manage all the connections
itself. The connection pool is initialized as soon as you execute the module.
You should never have to worry about connections in rethinkdbdash. Connections are created as they are needed, and in case of a host failure, the pool will try to open connections with an exponential back off algorithm.
The driver execute one query per connection. Now that rethinkdb/rethinkdb#3296 is solved, this behavior may be changed in the future.
Because the connection pool will keep some connections available, a script will not terminate. If you have finished executing your queries and want your Node.js script to exit, you need to drain the pool with:
r.getPoolMaster().drain();
The pool master by default will log all errors/new states on stderr
. If you do not
want to pollute stderr
, pass silent: true
when you import the driver and
provide your own log
method.
r = require('rethinkdbdash')({
silent: true,
log: function(message) {
console.log(message);
}
});
The pool is composed of a PoolMaster
that retrieve connections for n
pools where n
is the number of
servers the driver is connected to. Each pool is connected to a unique host.
To access the pool master, you can call the method r.getPoolMaster()
.
The pool emits a few events:
draining
: whendrain
is calledqueueing
: when a query is added/removed from the queue (queries waiting for a connection), the size of the queue is providedsize
: when the number of connections changes, the number of connections is providedavailable-size
: when the number of available connections changes, the number of available connections is provided
You can get the number of connections (opened or being opened).
r.getPoolMaster().getLength();
You can also get the number of available connections (idle connections, without a query running on it).
r.getPoolMaster().getAvailableLength();
You can also drain the pool as mentionned earlier with;
r.getPoolMaster().drain();
You can access all the pools with:
r.getPoolMaster().getPools();
The pool master emits the healthy
when its state change. Its state is defined as:
- healthy when at least one pool is healthy: Queries can be immediately executed or will be queued.
- not healthy when no pool is healthy: Queries will immediately fail.
A pool being healthy is it has at least one available connection, or it was just created and opening a connection hasn't failed.
r.getPoolMaster().on('healthy', function(healthy) {
if (healthy === true) {
console.log('We can run queries.');
}
else {
console.log('No queries can be run.');
}
});
If you do not wish to use rethinkdbdash connection pool, you can implement yours. The connections created with rethinkdbdash emits a "release" event when they receive an error, an atom, or the end (or full) sequence.
A connection can also emit a "timeout" event if the underlying connection times out.
Rethinkdbdash automatically coerce cursors to arrays. If you need a raw cursor,
you can call the run
command with the option {cursor: true}
or import the
driver with {cursor: true}
.
r.expr([1, 2, 3]).run().then(function(result) {
console.log(JSON.stringify(result)) // print [1, 2, 3]
})
r.expr([1, 2, 3]).run({cursor: true}).then(function(cursor) {
cursor.toArray().then(function(result) {
console.log(JSON.stringify(result)) // print [1, 2, 3]
});
})
Note: If a query returns a cursor, the connection will not be released as long as the cursor hasn't fetched everything or has been closed.
Readable streams can be
synchronously returned with the toStream([connection])
method.
var fs = require('fs');
var file = fs.createWriteStream('file.txt');
var r = require('rethinkdbdash')();
r.table('users').toStream()
.on('error', console.log)
.pipe(file)
.on('error', console.log)
.on('end', function() {
r.getPool().drain();
});
Note: The stream will emit an error if you provide it with a single value (streams, arrays and grouped data work fine).
Note: null
values are currently dropped from streams.
You can create a Writable
or Transform streams by
calling toStream([connection, ]{writable: true})
or
toStream([connection, ]{transform: true})
on a table.
By default, a transform stream will return the saved documents. You can return the primary
key of the new document by passing the option format: 'primaryKey'
.
This makes a convenient way to dump a file your database.
var file = fs.createReadStream('users.json')
var table = r.table('users').toStream({writable: true});
file.pipe(transformer) // transformer would be a Transform stream that splits per line and call JSON.parse
.pipe(table)
.on('finish', function() {
console.log('Done');
r.getPool().drain();
});
The then
and catch
methods are implemented on a Term
- returned by any methods
like filter
, update
etc. They are shortcut for this.run().then(callback)
and
this.run().catch(callback)
.
This means that you can yield
any query without calling run.
var bluebird = require('bluebird');
var r = require('rethinkdbdash')();
bluebird.coroutine(function*() {
try {
var result = yield r.table('users').get('orphee@gmail.com').update({name: 'Michel'});
assert.equal(result.errors, 0);
} catch(err) {
console.log(err);
}
});
Note: You have to start Node >= 0.11 with the --harmony
flag.
You can set the maximum nesting level and maximum array length on all your queries with:
r.setNestingLevel(<number>)
r.setArrayLimit(<number>)
Rethinkdbdash will ignore the keys/values where the value is undefined
instead
of throwing an error like the official driver.
If your query fails, the driver will return an error with a backtrace; your query will be printed and the broken part will be highlighted.
Backtraces in rethinkdbdash are tested and properly formatted. Typically, long backtraces are split on multiple lines and if the driver cannot serialize the query, it will provide a better location of the error.
The server may return confusing error messages when the wrong number of arguments is provided (See rethinkdb/rethinkdb#2463 to track progress). Rethinkdbdash tries to make up for it by catching errors before sending the query to the server if possible.
The tree representation of the query is built step by step and stored which avoid recomputing it if the query is re-run.
The code was partially optimized for v8, and is written in pure JavaScript which avoids errors like issue #2839
Update test/config.js
if your RethinkDB instance doesn't run on the default parameters.
Make sure you run a version of Node that supports async/await
and run:
npm test
Longer tests for the pool:
mocha long_test/discovery.js -t 50000
mocha long_test/static.js -t 50000
-
Why rethinkdbdash?
Rethinkdbdash was built as an experiment for promises and a connection pool. Its purpose was to test new features and improve the official driver. Today, rethinkdbdash still tries to make the developer experience as pleasant as possible - like with the recent support for Node.js streams.
Some features like promises have been back ported to the official driver, some like the connection pool and streams are on their way.
-
Is it stable?
Yes. Rethinkdbdash is used by quite many people. The driver is also used by
thinky
, and has been and is still being tested in the wild. -
Does it work with io.js?
All the tests pass with io.js so yes.
-
Is rethinkdbdash going to become the JavaScript official driver?
Not (yet?), maybe :)
Completely replacing the driver requires some work:
- Integrate the driver in RethinkDB suite test.
- Support HTTP connections.
- Rollback some default like the coercion of cursors to arrays.
-
Can I contribute?
Feel free to send a pull request. If you want to implement a new feature, please open an issue first, especially if it's a non backward compatible one.
To build the browser version of rethinkdbdash, run:
node browserify.js