Skip to content


Subversion checkout URL

You can clone with
Download ZIP
MongoDB and GridFS support for Mule 3.

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.


The MongoDB transport allows you to insert and retrieve data from MongoDB collections and buckets.

A MongoDB inbound-endpoint accepts a JSON query that is used to poll a MongoDB collection or bucket at the specified interval for results. The resultant message can then be processed by a component or routed to an outbound-endpoint.

The MongoDB outbound-endpoint inserts either Map or JSON data into the specified collection. Files, streams and byte arrays can be persisted as BSON data using the inbound-endpoint with a bucket. See the documentation on GridFS below for details.

Please open a JIRA if you experience any issues. For general comments, suggestions, etc I can be reached at


Add the following to your project's pom.xml:



The following examples will demonstrate using MongoDB inbound and outbound endpoints in conjunction with the STDIO transport. Note that the configs omit the namespace declarations for clarity.


    <mongodb:connector name="mongodbConnector"

   <stdio:connector name="stdioConnector" 
             promptMessage="Enter some JSON"/>

        <service name="MongoDB Test Input Service">
                <stdio:inbound-endpoint system="IN"/>
                    <mongodb:outbound-endpoint collection="stuff"/>

Starting Mule will produce a prompt to enter some JSON which will be inserted
into the "stuff" collection belonging to the "mule-mongodb" database:

Enter some JSON {"hello": "world"} INFO 2010-06-28 22:24:03,339 [mongodbConnector.dispatcher.1] org.mule.transport.mongodb.MongoDBMessageDispatcher: Connected: endpoint.outbound.mongodb://stuff

You can also dispatch Map payloads to insert into MongoDB. The following
illustrates this using a nested Map with MuleClient in a Groovy script:

def client = new MuleClient();
def person = [name: "Johnny Five", address: [city: "Brooklyn", state: "NY"]]
client.dispatch("mongodb://stuff", person, null);

Updating Documents

Documents can also be updated with an outbound-endpoint. In order to update data you need to set the "dispatch_mode" property on the outbound-message to "update". If the payload of the message contains an "_id" key then the transport will attempt to update a document with that ID. If the document doesn't exist then it will be inserted using the ID. To specify the object to update without supplying an ID you can set the "update_query" with a JSON query to identify the objects you wish to update. Here's an example:

def client = new MuleClient();
def person = [address: [city: "Queens", state: "NY"]]
client.dispatch("mongodb://stuff", person, [dispatch_mode: "update",
update_query:"\{"name": "Johnny Five"\}"]);

The update will fail if the query doesn't return any documents. To instead have the dispatcher create documents when they don't exist you can set the "update_upsert" property on the outbound message:

client.dispatch("mongodb://stuff", person, [dispatch_mode: "update",
update_query:"\{"name": "Johnny Five"\}"], update_upsert:"true");

By default the dispatcher will only update the first document returned by the query. To have it update all the documents you can set the "update_multi" property on the outbound message:

client.dispatch("mongodb://stuff", person, [dispatch_mode: "update",
update_query:"\{"name": "Johnny Five"\}"], update_multi:"true");

Deleting Documents

Documents can be deleted by setting the "dispatch_mode" property of the outbound-message to "delete". The payload of the message will be used to construct the deletion query. For example:

client.send("mongodb://stuff", "{\"name\": \"Foo\"}", [dispatch_mode: "delete"];

Write Concerns

Write concerns are now supported by this transport. The desired WriteConcern can be set on the "write_concern" property on a message. Allowed values are NONE,NORMAL,SAFE,FSYNC_SAFE, and REPLICAS_SAFE. The default is NONE.

Retrieving Data

The MongoDB transport allows you to poll a collection using a query at a specified interval and send the results through to a component or outbound endpoint. The following demonstrates a query that returns all objects in a collection with a timestamp greater then January 1, 2010:

        <service name="MongoDB Polling Service">
                <mongodb:inbound-endpoint collection="stuff" 
                              query='{ "ts": { "$gt": 1262304000000}  }'/>
                    <stdio:outbound-endpoint system="OUT"/>

Documents matching the query will be sent to the outbound endpoint. For instance:

INFO  2010-06-29 16:07:22,738 [connector.stdio.0.dispatcher.1] 
org.mule.transport.stdio.StdioMessageDispatcher: Connected: 
[{ "_id" : { "$oid" : "4c2a513d42157f56a3d82a9c"} , "x" : 203 , "y" : 102 , "ts"

Note that the payload of these messages are instances of BasicDBObject from the MongoDB driver. These seem to serialize to JSON when toString() is invoked, hence the above output. This also means you can manipulate the output however you see fit, either by interacting with BasicDBObject or serializing to JSON via toString().

Retrieving Data with an outbound-endpoint

Queries are also supported on outbound-endpoints. This is useful in flows, here's an example:

    <flow name="Query Test With Expression">
        <vm:inbound-endpoint path="">
                <add-message-property key="FOO" value="#[header:INBOUND:FOO]"/>
        <mongodb:outbound-endpoint collection="query" query='{ "name":
"#[header:FOO]" }'
        <vm:outbound-endpoint path="query.expr.out"/>

This demonstrates a flow that performs a query against a collection using a expression to enrich the query.

Using GridFS

As of version, the MongoDB transport supports GridFS.

Inserting into a Bucket

The following demonstrates how to use the file transport to insert files from a directory into a GridFS bucket.

        <service name="MongoDB Test Input Service">

You set the "filename" header of the message to set the filename of the object in MongoDB. This is probably a good thing to set when inserting streamed data or byte arrays.

Polling a Bucket

The following demonstrates how to poll a GridFS bucket for files and, using the file transport, output them to a directory.

        <service name="MongoDB Polling Service">
                    query='{ "filename": "*.dat" }' 
                    <file:outbound-endpoint path="/tmp/foo">

Note that the payloads of objects coming back from GridFS will be a collection of instances of GridFSDBFile. You can use the provided mongodb:db-file-to-byte-array or mongodb:db-file-to-input-stream to automatically extract either the byte array or stream from the GridFSDBFile. These transformers will additionally propagate the "filename" property of the GridFSDBFile to the returned message, allowing the files to be named predictably without any additional effort.

Getting a Single File

You can use the gridfs-file message processor to retrieve a single file from GridFS. This is typically useful in a flow, for example:

<flow name="GridFS Load File">
    <vm:inbound-endpoint path=""/>
    <mongodb:gridfs-file bucket="foo"/>
    <vm:outbound-endpoint path="gridfs.load.out"/>

By default gridfs-file will use the payload of the message as the filename to load from GridFS. An expression can be used to control how the payload is evaluated by setting the "fileExpression" property on the processor. You can also avoid using the message payload as the filename altogether by setting the "query" property on the processor, for instance:

<flow name="GridFS Query For File By ID">
    <vm:inbound-endpoint path="">
            <add-message-property key="MONGO_ID"
    <mongodb:gridfs-file bucket="foo" query='{ "_id": "#[header:MONGO_ID]" }' />
    <vm:outbound-endpoint path="gridfs.query.out"/>

Using a Requestor

You can use MuleClient to fetch GridFSDBFile from MongoDB:

MuleClient client = new MuleClient();
List results = (List) client.request("mongodb://bucket:files?query=" +
URLEncoder.encode("{\"filename\":\"foo\"}","UTF-8"), 15000).getPayload();

Note that the JSON query needs to encoded. The results, as with the inbound-endpoint, will be instances of GridFSDBFile.

The objectId Property

Messages sent to a collection or bucket will get an outbound property called "objectId" set corresponding to the "_id" field of the document or file. You can reference this property when chaining mongodb endpoints in a chaining-router or flow. This is primarily useful in creating subsequent DBRef's in the processing chain.

Something went wrong with that request. Please try again.