roachjs

RoachJS - CockroachDB Driver

• Introduction
• Installation
• Documentation
• CockroachDB

Introduction

This client is a port from the original Golang client. Internally it's is more or less the same, but this driver provides a friendlier javascript interface.

Installation

$ npm install roachjs

Documentation

• Examples
  • Initiating a client
  • Basic client usage I
  • Advanced client usage I (Prepare & Flush)
  • Advanced client usage II (Transactions)
• Interface
  • new Client(opts)
    • .get(key, callback)
    • .put(key, value, callback)
    • .conditionalPut(key, value, ifValue, callback)
    • .increment(key, increment, callback)
    • .contains(key, callback)
    • .scan(start_key, end_key, limit, callback)
    • .delete(key, callback)
    • .deleteRange(start_key, end_key, limit, callback)
    • .prepare()
      • .flush(callback)
    • .runTransaction(opts, transaction, callback)
• Extra
  • Response properties
  • Transaction function
  • Compiling .proto files
  • Custom clock module

Examples

Initiating a client

var Roach = require('roachjs')

var client = new Roach({
    uri: 'https://localhost:8080'
})

module.exports = client

Basic client usage I

client.get("sample_key", function(err, value, res) {
    if(err) throw err

    client.put("other_key", value, function(err, res) {
        if(err) {
            // Failed
        }
        else {
            // Sucess
        }
    })
})

Advanced client usage I (Prepare & Flush)

// You should prepare your queries and send them in a single batch
// For optimal performance
var c = client.prepare()

// This callback will be the first to be executed
c.get("sample_key", function(err, value, res) {
    if(err) throw err

    // Do something...
})

c.get("sample_key2", function(err, value, res) {
    if(err) throw err

    // Do something...
})

c.put("some_key", "some_value", function(err) {
    if(err) throw err

    // Do something
})

// The flush callback is the last one to be called
c.flush(function(err, res) {
    if(err) throw err

    console.log('Sucessfuly flushed %d queries.', res.responses.length)
})

Advanced client usage II (Transactions)

var opts = {
    name: "transaction example",
}

var errNoApples = new Error('Insufficient apples!')

var transaction = function(txn, commit, abort) {
    txn.get("applesInStock", function(err, value, res) {
        if(err || applesInStock.err) {
                return abort(err)
            }

            var dispatch = 5
            var inStock = parseInt(applesInStock.value)

            if(inStock < dispatch) {
                return abort(errNoApples)
        }

        // Upgrade for a prepared client
        txn = txn.prepare()

        txn.increment("applesInStock", -dispatch)
        txn.increment("applesInRoute", +dispatch)

        // Commit automatically flushes
        commit()
    })
}

client.runTransaction(opts, transaction, function(err, res) {
    if(err === errNoApples) {
        // Alert user there are no more apples...
    }
    else if(err) {
        // Transaction failed...
    }
    else {
        // Transaction commited...
    }
})

Interface

new Client(opts)

Returns a new roachjs client with options.

Parameters

name	type	description
opts	object	see

Client options

opt	description	default
uri	uri to the cockroach http endpoint	https://localhost:8080/
host	host or ip to the cockroach http endpoint	localhost
port	port to the cockroach http endpoint	8080
ssl	connect throught https	true
user	user to run the requests with	root
retry	retry requests when cockroach responds with a busy signal	true
http	http module to use	require('https')
agent	http agent to use on the requests (read more)	new http.Agent()
clock	clock module to use (read more)	internal clock module

Methods

method
get
put
conditionalPut
contains
increment
scan
delete
deleteRange
prepare
runTransaction

client.get(key, callback)

Gets a single entry from the datastore, specified by key.

Parameters

name	type	description
key	string
callback	callback	function(err, value, res) {}

Callback

name	type	description
err	Error()
value	Buffer
res	object	see

Example

client.get("key", function(err, value, res) {})

client.put(key, value, callback)

Puts a value in the datastore in the specified key. Ideally you should send in buffers, but you can pass a string, preferably an utf-8 encoded string.

Parameters

name	type	description
key	string
value	Buffer, string
callback	callback	function(err, res) {}

Callback

name	type	description
err	Error()
res	object	see

Example

client.put("key", "value", function(err, res) {})

client.conditionalPut(key, value, ifValue, callback)

ConditionalPut sets the value for a key if the existing value matches the ifValue. Specifying an empty or null ifValue means the entry must not yet exist.

Parameters

name	type	description
key	string
value	Buffer, string
ifValue	Buffer, string, null	use null to put if entry doens't exists
callback	callback	function(err, res) {}

Callback

name	type	description
err	Error()
actualValue	Buffer	If conditional put fails this value is set
res	object	see

Example

client.conditionalPut("status", "running", "stopped", function(err, actualValue, res) {})
client.conditionalPut("status", "new", null, function(err, actualValue, res) {})

client.contains(key, callback)

Contains determines if a key exists in the datastore.

Parameters

name	type	description
key	string
callback	callback	function(err, exists, res) {}

Callback

name	type	description
err	Error()
exists	boolean
res	object	see

Example

client.contains("john", function(err, exists, res) {
    if(exists === true) {
        // john exists in the datastore
    }
})

client.increment(key, increment, callback)

Increment increments the value at the specified key by some increment value. Once called for a key, Put & Get will return errors; only Increment will continue to be a valid command. The value must be deleted before it can be reset using Put.

Parameters

name	type	description
key	string
increment	integer
callback	callback	function(err, newValue, res) {}

Callback

name	type	description
err	Error()
newValue	integer	the new value for this counter, after the increment operation
res	object	see

Example

client.increment("counter", 5, function(err, newValue, res) {
    console.log('counter current value is', newValue)
})

client.scan(start_key, end_key, limit, callback)

Scan the datastore for keys in the range of the start_key and end_key, limiting the result by limit.

Parameters

name	type	description
key	string
start_key	string
end_key	string
limit	integer
callback	callback	function(err, rows, res) {}

Callback

name	type	description
err	Error()
rows	array
res	object	see

Example

client.scan("a", "Z", 100, function(err, rows, res) {
    for(row as rows) {
        console.log(row)
    }
})

client.delete(key, callback)

Delete an entry from the datastore specified by key.

Parameters

name	type	description
key	string
callback	callback	function(err, res) {}

Callback

name	type	description
err	Error()
res	object	see

Example

client.delete("key", function(err, res) {})

client.deleteRange(start_key, end_key, limit, callback)

Delete all keys found in a range, from start_key to end_key, limited by limit.

Parameters

name	type	description
key	string
start_key	string
end_key	string
limit	integer
callback	callback	function(err, deleted, res) {}

Callback

name	type	description
err	Error()
deleted	integer	number of entries deleted
res	object	see

Example

client.deleteRange("a", "Z", 100, function(err, deleted, res) {
    console.log('deleted %d entries', deleted)
})

client.prepare()

Return you a new prepared client. It has all the methods from the original client. Read Advanced client usage II to understand how to use this client. You should always use this client when sending in multiple queries, this will batch them together in a single request.

Methods

method	description
flush	Flush the prepared queries

Example

var c = client.prepare()

c.get("key", function(err, value, res) {
    // Do something...
})

c.get("key2", function(err, value, res) {
    // Do something...
})

c.put("key3", "value", function(err, res) {
    // Do something...
})

c.flush()

client.flush(callback)

Flush the prepared queries buffer, and send it as a batch request.

Parameters

name	type	description
callback	callback	optional

Callback

name	type	description
err	Error()	batch request failed
res	object	see

Example

client.flush(function(err, res) {
    if(err) {
        // Flush failed..
    }
    else {
        console.log('flushed %d queries.', res.responses.length)
    }
})

Returns

Returns an response object.

property	type	description
err	Error()	is null if no error was returned
value	string, number, boolean	general response value
res	object	see

client.runTransaction(opts, transaction, callback)

RunTransaction executes a retryable transaction function in the context of a distributed transaction. The transaction is automatically aborted if retryable function returns any error aside from recoverable internal errors, and is automatically committed otherwise. retryable should have no side effects which could cause problems in the event it must be run more than once. The opts contains transaction settings.

Parameters

name	type	description
opts	object	options
transacation	retryable function	function(txn, commit, abort) {}
callback	callback	function(err, res) {}

Transaction options

opt	description	default
name	transaction name for debugging	""
isolation		0

Callback

name	type	description
err	Error()	if transaction fails
res	object	see

Extra

Response properties

The res argument contains the full database response, each database command can contain a different set of properties. This document will try to state some of the possible properties.

Properties

property	type	description
timestamp	integer	timestamp of the returned entry
wall_time	integer	timestamp of when the read or write operation was performed

Transaction function

The transaction function is an retryable function, it may be executed more than once. This function should never forget to call commit or abort. Throwing an error inside this function also aborts the transaction.

Arguments

name	type	description
txn	Prepared client	this client is the same as client.prepare(), you can flush yourself if you don't wan't to commit yet.
commit	callback	to try to commit transaction
abort	callback	to abort transaction

• abort() accepts an optional Error. This error will be passed to the .runTransaction callback.

Example

var transaction = function(txn, commit, abort) {
    txn = txn.prepare()

    for(var i = 0; i < 100; i++) {
        var key = i.toString()

        txn.put(key, "hello")
    }

    // Commit automatically flushes
    // the prepared transaction.
    commit()
}

Compiling .proto files

Cockroachdb's protocol buffer files are mantained at a repository called cockroachdb/cockroach-proto, this is maintained as a subtree in this library, in case you need to manually update or change them, follow this steps.

If you want to sync them with the latest proto files

You will need to update the folder cockroach-proto with the latest content of the cockroachdb/cockroach-proto repository, you could do this with:

$ git subtree pull -P cockroach-proto [e-mail]:cockroachdb/cockroach-proto.git master --squash

• Notice: I'm not sure if this is a good pattern, just be sure to update the folder contents.

Recompile the .proto files

Run the following npm script to compile the .proto files to javascript, it will automatically place the files in the lib folder.

$ npm run build-proto

Custom clock module

You may wan't to use a custom clock module in some cases, you can pass it through the clock option, when instantiating a new Client. Your clock module should have a now method, and this method should return the current timestamp in nanoseconds, here is an example:

var clockModule = module.exports = {
    now: function TimestampNanoseconds() {
        // Should return the current timestamp in nanoseconds
        return ...;
    }
}

var client = new Client({
    clock: clockModule
})

Maintainers

• herenow