Mongolastic
Mongolastic enables you to migrate your datasets from a mongod node to an elasticsearch node and vice versa. Since mongo and elastic servers can run with different characteristics, the tool provides several optional and required features to ably connect them. Mongolastic works with a yaml or json configuration file to begin a migration process. It reads your demand on the file and start syncing data in the specified direction.
How it works
First, you can either pull the corresponding image of the app from Docker Hub
Supported tags and respective Dockerfile links:
-
1.4,1.4.4,latest(master/Dockerfile) -
1.4.3(1.4.2/Dockerfile)
or download the latest mongolastic.jar file.
Second, create a yaml or json file which must contain the following structure:
misc:
dindex:
name: <string> (1)
as: <string> (2)
ctype:
name: <string> (3)
as: <string> (4)
direction: (em | me) (5)
batch: <number> (6)
dropDataset: <bool> (7)
mongo:
host: <ip-address> (8)
port: <number> (9)
query: "mongo-query" (10)
project: "projection" (11)
auth: (12)
user: <string>
pwd: "password"
source: <db-name>
mechanism: ( plain | scram-sha-1 | x509 | gssapi | cr )
elastic:
host: <ip-address> (13)
port: <number> (14)
dateFormat: "<format>" (15)
longToString: <bool> (16)
clusterName: <string> (17)
auth: (18)
user: <string>
pwd: "password"-
the database/index name to connect to.
-
another database/index name in which documents will be located in the target service (Optional)
-
the collection/type name to export.
-
another collection/type name in which indexed/collected documents will reside in the target service (Optional)
-
direction of the data transfer. the default direction is me (that is, mongo to elasticsearch). You can skip this option if your data move from mongo to es.
-
Override the default batch size which is normally 200. (Optional)
-
configures whether or not the target table should be dropped prior to loading data. Default value is true (Optional)
-
the name of the host machine where the
mongodis running. -
the port where the
mongodinstance is listening. -
data will be transferred based on a json mongodb query (Optional)
-
with 1.4.1, you can manipulate documents that will be migrated from mongo to es based on the
$projectoperator (Optional) -
as of v1.3.5, you can access an auth mongodb by giving auth configuration. (Optional)
-
the name of the host machine where the
elastic nodeis running. -
the transport port where the transport module will communicate with the running elastic node. E.g. 9300 for node-to-node communication.
-
a custom formatter for Date fields rather than the default DateCodec (Optional)
-
serialize long value as a string for backwards compatibility with other tools (Optional)
-
connect to a spesific elastic cluster (Optional)
-
as of v1.3.9, you can access an auth elastic search by giving auth configuration. (Optional)
Alternatively, a JSON file can be specified as a mongolastic configuration file including the same YAML file structure above.
{
"misc": {
"dindex": {
"name": "twitter",
"as": "media"
},
"ctype": {
"name": "tweets",
"as": "posts"
},
"direction": "me",
"batch": 400,
"dropDataset": true
},
"mongo": {
"host": "127.0.0.1",
"port": 27017,
"query": "{ lang: 'en' }",
"project": "{ user:1, name:'$user.name', location: { $substr: [ '$user.location', 10, 15 ] }}",
"auth": {
"user": "joe",
"pwd": "1234",
"source": "twitter",
"mechanism": "scram-sha-1"
}
},
"elastic": {
"host": "127.0.0.1",
"port": 9300,
"dateFormat": "yyyy-MM-dd",
"longToString": true,
"auth": {
"user": "joe",
"pwd": "4321"
}
}
}Example #1
The following files have the same configuration details:
misc:
dindex:
name: twitter
as: kodcu
ctype:
name: tweets
as: posts
mongo:
host: localhost
port: 27017
query: "{ 'user.name' : 'kodcu.com'}"
elastic:
host: localhost
port: 9300{
"misc": {
"dindex": {
"name": "twitter",
"as": "kodcu"
},
"ctype": {
"name": "tweets",
"as": "posts"
}
},
"mongo": {
"host": "localhost",
"port": 27017,
"query": "{ 'user.name' : 'kodcu.com'}"
},
"elastic": {
"host": "localhost",
"port": 9300
}
}the config says that the transfer direction is from mongodb to elasticsearch, mongolastic first looks at the tweets collection, where the user name is kodcu.com, of the twitter database located on a mongod server running on default host interface and port number. If It finds the corresponding data, It will start copying those into an elasticsearch environment running on default host and transport number. After all, you should see a type called "posts" in an index called "kodcu" in the current elastic node. Why the index and type are different is because "dindex.as" and "ctype.as" options were set, these indicates that your data being transferred exist in posts type of the kodcu index.
After downloading the jar or pulling the image and providing a conf file, you can either run the tool as:
$ java -jar mongolastic.jar -f config.file
or
$ docker run --rm -v $(PWD)/config.file:/config.file --net host ozlerhakan/mongolastic:<tag> config.file
Example #2
Using the project field, you are able to manipulate documents when migrating them from mongodb to elasticsearch. For more examples about the $project operator of the aggregation pipeline, take a look at its documentation.
misc:
dindex:
name: twitter
ctype:
name: tweets
mongo:
host: 192.168.10.151
port: 27017
project: "{ user: 1, name: '$user.name', location: { $substr: [ '$user.location', 10, 15 ] }}" (1)
elastic:
host: 192.168.10.152
port: 9300-
the migrated documents will include the user field and contain new fields
nameandlocation.
|
Note
|
Every attempt of running the tool drops the mentioned db/index in the target environment unless the dropDataset parameter is configured otherwise. |
License
Mongolastic is released under MIT.