A database with multi master replication and merge support based on leveldb, fwdb and scuttleup
npm install multi-master-merge
var mmm = require('multi-master-merge')
var level = require('level')
var mdb = mmm(level('data.db'), {encoding:'utf-8'})
mdb.put('hello', 'world', function(err, doc) {
console.log('Inserted:', doc)
mdb.get('hello', function(err, docs) {
console.log('"hello" contains the following:', docs)
})
})
When you do mdb.get(key, cb)
you will always get an array of documents back.
The reason for this is to support multi master replication which means
that we might have multiple values for a given key.
To replicate your database simply open a sync
stream and pipe it to another
database
// a and b are two database instances
var s1 = a.sync() // open a sync stream for db a
var s2 = b.sync() // open a sync stream for db b
s1.pipe(s2).pipe(s1) // pipe them together to start replicating
Updates will now be replicated between the two instances.
If two databases inserts a document on the same key both of them will be
present if you do a mdb.get(key)
a.put('hello', 'a')
b.put('hello', 'b')
setTimeout(function() {
a.get('hello', function(err, docs) {
console.log(docs) // will print [{peer:..., value:'a'}, {peer:..., value:'b'}]
})
}, 1000) // wait a bit for the inserts to replicate
To combine multiple documents into a single one use mdb.merge(key, docs, newValue)
If we consider the above replication scenario we have two documents for the key hello
a.get('hello', function(err, docs) {
console.log(docs) // will print [{peer:..., value:'a'}, {peer:..., value:'b'}]
})
To merge them into a single document do
a.get('hello', function(err, docs) {
a.merge('hello', docs, 'a + b')
})
Merges will replicate as well
// wait a bit for a to replicate to b
b.get('hello', function(err, docs) {
console.log(docs) // will print [{peer:..., value:'a + b'}]
})
Create a new database. db
is a levelup instance.
Options can include
{
id: peerId, // defaults to a cuid
encoding: 'json' | 'utf-8' | 'binary', // the value encoding. defaults to binary
preupdate: function(logData, cb) {}, // set a preupdate hook
postupdate: function(logData, cb) {} // set a postupdate hook
}
Insert a new document. Callback is called with cb(err, doc)
where doc is the inserted document.
Get documents stored on key
. Callback is called with cb(err, docs)
.
Get a stream of all {key:key, peer:peer, seq:seq, value:docs}
pair in the database.
You can pass in gt
,gte
,lt
,lte
options similar to levelup.
Similar to createReadStream
but only returns keys
Get a stream of all {key:key, value:doc}
pairs in the log (all revisions of docs). Works about 10x faster than createValueStream, but returns all revisions, even merged ones.
Merge multiple documents into a new document. Callback is called with cb(err, doc)
where doc is the inserted merged document.
Returns a replication stream that can be piped to another mdb
instance to replicate between them.
Per defaults changes are replicated both ways. If you only want to push changes to another instance set
{mode: 'push'}
and if you only want to get changes do {mode: 'pull'}
The used fwdb instance
The used scuttleup instance
MIT