Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Dgraph Client for Java

Build Status Coverage Status

A minimal implementation for a Dgraph client for Java 1.8 and above, using grpc.

This client follows the Dgraph Go client closely.

Before using this client, we highly recommend that you go through, and understand how to run and work with Dgraph.

Use Discuss Issues for reporting issues about this repository.

Table of Contents


grab via Maven:


or Gradle:

compile 'io.dgraph:dgraph4j:20.11.0'

Supported Versions

Depending on the version of Dgraph that you are connecting to, you will have to use a different version of this client.

Dgraph version dgraph4j version java version
1.0.X 1.X.X 1.9.X
1.1.0 - 2.X.X 2.X.X 1.9.X
>= 20.XX.X 20.XX.X 1.9.X

Note regarding Java 1.8.x support:

  • If you aren't using gRPC with TLS, then the above version table will work for you with Java 1.8.x too.

  • If you're using gRPC with TLS on Java 1.8.x, then you will need to follow gRPC docs here . Basically, it will require you to add the following dependency in your app with correct version for the corresponding grpc-netty version used by dgraph4j. You can find out the correct version of the dependency to use from the version combination table in this section in grpc-netty docs.

    For maven:

      <version><!-- See table in gRPC docs for correct version --></version>

    For Gradle:

    compile 'io.netty:netty-tcnative-boringssl-static:<See table in gRPC docs for correct version>'

    The following table lists the grpc-netty versions used by different dgraph4j versions over time, along with the supported versions of netty-tcnative-boringssl-static for the corresponding grpc-netty version:

    dgraph4j version grpc-netty version netty-tcnative-boringssl-static version
    1.0.0-1.2.0 1.7.0 2.0.6.Final
    1.4.0-1.6.0 1.10.0 2.0.7.Final
    1.7.0 1.15.0 2.0.12.Final
    1.7.3-1.7.5 1.15.1 2.0.12.Final
    2.0.0-2.1.0 1.22.1 2.0.25.Final
    20.03.0-20.03.3 1.26.0 2.0.26.Final
    >= 20.11.0 1.34.1 2.0.31.Final

    So, for example, if you were using dgraph4j v20.03.0, then you would need to use 2.0.26-Final as the version for netty-tcnative-boringssl-static dependency as suggested by gRPC docs for grpc-netty v1.26.0.


Build and run the DgraphJavaSample project in the samples folder, which contains an end-to-end example of using the Dgraph Java client. Follow the instructions in the README of that project.


This library supports two styles of clients, the synchronous client DgraphClient and the async client DgraphAsyncClient. A DgraphClient or DgraphAsyncClient can be initialised by passing it a list of DgraphBlockingStub clients. The anyClient() API can randomly pick a stub, which can then be used for GRPC operations. In the next section, we will explain how to create a synchronous client and use it to mutate or query dgraph. For the async client, more details can be found in the Using the Asynchronous Client section.

Using the Synchronous Client

Creating a Client

The following code snippet shows how to create a synchronous client using three connections.

ManagedChannel channel1 = ManagedChannelBuilder
    .forAddress("localhost", 9080)
DgraphStub stub1 = DgraphGrpc.newStub(channel1);

ManagedChannel channel2 = ManagedChannelBuilder
    .forAddress("localhost", 9082)
DgraphStub stub2 = DgraphGrpc.newStub(channel2);

ManagedChannel channel3 = ManagedChannelBuilder
    .forAddress("localhost", 9083)
DgraphStub stub3 = DgraphGrpc.newStub(channel3);

DgraphClient dgraphClient = new DgraphClient(stub1, stub2, stub3);

Creating a Client for Slash Graphql Endpoint

If you want to connect to Dgraph running on your Slash GraphQL instance, then all you need is the URL of your Slash GraphQL endpoint and the API key. You can get a client using them as follows :

DgraphStub stub = DgraphClient.clientStubFromSlashEndpoint("", "your-api-key");
DgraphClient dgraphClient = new DgraphClient(stub);

Creating a Secure Client using TLS

To setup a client using TLS, you could use the following code snippet. The server needs to be setup using the instructions provided here.

If you are doing client verification, you need to convert the client key from PKCS#1 format to PKCS#8 format. By default, grpc doesn't support reading PKCS#1 format keys. To convert the format, you could use the openssl tool.

First, let's install the openssl tool:

apt install openssl

Now, use the following command to convert the key:

openssl pkcs8 -in -topk8 -nocrypt -out

Now, you can use the following code snippet to connect to Alpha over TLS:

SslContextBuilder builder = GrpcSslContexts.forClient();
builder.trustManager(new File("<path to ca.crt>"));
// Skip the next line if you are not performing client verification.
builder.keyManager(new File("<path to>"), new File("<path to>"));
SslContext sslContext =;

ManagedChannel channel = NettyChannelBuilder.forAddress("localhost", 9080)
DgraphGrpc.DgraphStub stub = DgraphGrpc.newStub(channel);
DgraphClient dgraphClient = new DgraphClient(stub);

Check Dgraph version

Checking the version of the Dgraph server this client is interacting with is as easy as:

Version v = dgraphClient.checkVersion();

Checking the version, before doing anything else can be used as a test to find out if the client is able to communicate with the Dgraph server. This will also help reduce the latency of the first query/mutation which results from some dynamic library loading and linking that happens in JVM (see this issue for more details).

Login Using ACL

dgraphClient.login(USER_ID, USER_PASSWORD);

Altering the Database

To set the schema, create an Operation object, set the schema and pass it to DgraphClient#alter method.

String schema = "name: string @index(exact) .";
Operation operation = Operation.newBuilder().setSchema(schema).build();

Starting Dgraph version 20.03.0, indexes can be computed in the background. You can call the function setRunInBackground(true) as shown below before calling alter. You can find more details here.

String schema = "name: string @index(exact) .";
Operation operation = Operation.newBuilder()

Operation contains other fields as well, including drop predicate and drop all. Drop all is useful if you wish to discard all the data, and start from a clean slate, without bringing the instance down.

// Drop all data including schema from the dgraph instance. This is useful
// for small examples such as this, since it puts dgraph into a clean
// state.

Creating a Transaction

There are two types of transactions in dgraph, i.e. the read-only transactions that only include queries and the transactions that change data in dgraph with mutate operations. Both the synchronous client DgraphClient and the async client DgraphAsyncClient support the two types of transactions by providing the newTransaction and the newReadOnlyTransaction APIs. Creating a transaction is a local operation and incurs no network overhead.

In most of the cases, the normal read-write transactions is used, which can have any number of query or mutate operations. However, if a transaction only has queries, you might benefit from a read-only transaction, which can share the same read timestamp across multiple such read-only transactions and can result in lower latencies.

For normal read-write transactions, it is a good practise to call Transaction#discard() in a finally block after running the transaction. Calling Transaction#discard() after Transaction#commit() is a no-op and you can call discard() multiple times with no additional side-effects.

Transaction txn = dgraphClient.newTransaction();
try {
    // Do something here
    // ...
} finally {

For read-only transactions, there is no need to call Transaction.discard, which is equivalent to a no-op.

Transaction readOnlyTxn = dgraphClient.newReadOnlyTransaction();

Read-only transactions can be set as best-effort. Best-effort queries relax the requirement of linearizable reads. This is useful when running queries that do not require a result from the latest timestamp.

Transaction bestEffortTxn = dgraphClient.newReadOnlyTransaction()

Running a Mutation

Transaction#mutate runs a mutation. It takes in a Mutation object, which provides two main ways to set data: JSON and RDF N-Quad. You can choose whichever way is convenient.

We're going to use JSON. First we define a Person class to represent a person. This data will be serialized into JSON.

class Person {
    String name
    Person() {}

Next, we initialise a Person object, serialize it and use it in Mutation object.

// Create data
Person person = new Person(); = "Alice";

// Serialize it
Gson gson = new Gson();
String json = gson.toJson(person);
// Run mutation
Mutation mu = Mutation.newBuilder()

Sometimes, you only want to commit mutation, without querying anything further. In such cases, you can use a CommitNow field in Mutation object to indicate that the mutation must be immediately committed.

Mutation can be run using the doRequest function as well.

Request request = Request.newBuilder()

Committing a Transaction

A transaction can be committed using the Transaction#commit() method. If your transaction consisted solely of calls to Transaction#query(), and no calls to Transaction#mutate(), then calling Transaction#commit() is not necessary.

An error will be returned if other transactions running concurrently modify the same data that was modified in this transaction. It is up to the user to retry transactions when they fail.

Transaction txn = dgraphClient.newTransaction();

try {
    // Perform any number of queries and mutations
    // and finally …
} catch (TxnConflictException ex) {
    // Retry or handle exception.
} finally {
    // Clean up. Calling this after txn.commit() is a no-op
    // and hence safe.

Running a Query

You can run a query by calling Transaction#query(). You will need to pass in a GraphQL+- query string, and a map (optional, could be empty) of any variables that you might want to set in the query.

The response would contain a JSON field, which has the JSON encoded result. You will need to decode it before you can do anything useful with it.

Let’s run the following query:

query all($a: string) {
  all(func: eq(name, $a)) {

First we must create a People class that will help us deserialize the JSON result:

class People {
    List<Person> all;
    People() {}

Then we run the query, deserialize the result and print it out:

// Query
String query =
"query all($a: string){\n" +
"  all(func: eq(name, $a)) {\n" +
"    name\n" +
"  }\n" +

Map<String, String> vars = Collections.singletonMap("$a", "Alice");
Response response = dgraphClient.newReadOnlyTransaction().queryWithVars(query, vars);

// Deserialize
People ppl = gson.fromJson(response.getJson().toStringUtf8(), People.class);

// Print results
System.out.printf("people found: %d\n", ppl.all.size());
ppl.all.forEach(person -> System.out.println(;

This should print:

people found: 1

You can also use doRequest function to run the query.

Request request = Request.newBuilder()

Running a Query with RDF response

You can get query results as an RDF response by calling either queryRDF() or queryRDFWithVars(). The response contains the getRdf() method, which will provide the RDF encoded output.

Note: If you are querying for uid values only, use a JSON format response

// Query
String query = "query me($a: string) { me(func: eq(name, $a)) { name }}";
Map<String, String> vars = Collections.singletonMap("$a", "Alice");
Response response =
    dgraphAsyncClient.newReadOnlyTransaction().queryRDFWithVars(query, vars).join();

// Print results

This should print (assuming Alice's uid is 0x2):

<0x2> <name> "Alice" .

Running an Upsert: Query + Mutation

The txn.doRequest function allows you to run upserts consisting of one query and one mutation. Variables can be defined in the query and used in the mutation. You could also use txn.doRequest to perform a query followed by a mutation.

To know more about upsert, we highly recommend going through the docs at

String query = "query {\n" +
  "user as var(func: eq(email, \"\"))\n" +
Mutation mu = Mutation.newBuilder()
    .setSetNquads(ByteString.copyFromUtf8("uid(user) <email> \"\" ."))
Request request = Request.newBuilder()

Running a Conditional Upsert

The upsert block also allows specifying a conditional mutation block using an @if directive. The mutation is executed only when the specified condition is true. If the condition is false, the mutation is silently ignored.

See more about Conditional Upsert Here.

String query = "query {\n" +
    "user as var(func: eq(email, \"\"))\n" +
Mutation mu = Mutation.newBuilder()
    .setSetNquads(ByteString.copyFromUtf8("uid(user) <email> \"\" ."))
    .setCond("@if(eq(len(user), 1))")
Request request = Request.newBuilder()

Setting Deadlines

It is recommended that you always set a deadline for each client call, after which the client terminates. This is in line with the recommendation for any gRPC client. Read this forum post for more details.

channel = ManagedChannelBuilder.forAddress("localhost", 9080).usePlaintext(true).build();
DgraphGrpc.DgraphStub stub = DgraphGrpc.newStub(channel);
ClientInterceptor timeoutInterceptor = new ClientInterceptor(){
    public <ReqT, RespT> ClientCall<ReqT, RespT> interceptCall(
            MethodDescriptor<ReqT, RespT> method, CallOptions callOptions, Channel next) {
        return next.newCall(method, callOptions.withDeadlineAfter(500, TimeUnit.MILLISECONDS));
DgraphClient dgraphClient = new DgraphClient(stub);

Setting Metadata Headers

Certain headers such as authentication tokens need to be set globally for all subsequent calls. Below is an example of setting a header with the name "auth-token":

// create the stub first
ManagedChannel channel =
ManagedChannelBuilder.forAddress(TEST_HOSTNAME, TEST_PORT).usePlaintext(true).build();
DgraphStub stub = DgraphGrpc.newStub(channel);

// use MetadataUtils to augment the stub with headers
Metadata metadata = new Metadata();
        Metadata.Key.of("auth-token", Metadata.ASCII_STRING_MARSHALLER), "the-auth-token-value");
stub = MetadataUtils.attachHeaders(stub, metadata);

// create the DgraphClient wrapper around the stub
DgraphClient dgraphClient = new DgraphClient(stub);

// trigger a RPC call using the DgraphClient

Helper Methods

Delete multiple edges

The example below uses the helper method Helpers#deleteEdges to delete multiple edges corresponding to predicates on a node with the given uid. The helper method takes an existing mutation, and returns a new mutation with the deletions applied.

Mutation mu = Mutation.newBuilder().build()
mu = Helpers.deleteEdges(mu, uid, "friends", "loc");

Closing the DB Connection

To disconnect from Dgraph, call ManagedChannel#shutdown on the gRPC channel object created when creating a Dgraph client.


Using the Asynchronous Client

Dgraph Client for Java also bundles an asynchronous API, which can be used by instantiating the DgraphAsyncClient class. The usage is almost exactly the same as the DgraphClient (show in previous section) class. The main differences is that the DgraphAsyncClient#newTransacation() returns an AsyncTransaction class. The API for AsyncTransaction is exactly Transaction. The only difference is that instead of returning the results directly, it returns immediately with a corresponding CompletableFuture<T> object. This object represents the computation which runs asynchronously to yield the result in the future. Read more about CompletableFuture<T> in the Java 8 documentation.

Here is the asynchronous version of the code above, which runs a query.

// Query
String query = 
"query all($a: string){\n" +
"  all(func: eq(name, $a)) {\n" +
"    name\n" +
 "}\n" +

Map<String, String> vars = Collections.singletonMap("$a", "Alice");

AsyncTransaction txn = dgraphAsyncClient.newTransaction();
txn.query(query).thenAccept(response -> {
    // Deserialize
    People ppl = gson.fromJson(res.getJson().toStringUtf8(), People.class);

    // Print results
    System.out.printf("people found: %d\n", ppl.all.size());
    ppl.all.forEach(person -> System.out.println(;

Checking the request latency

If you would like to see the latency for either a mutation or query request, the latency field in the returned result can be helpful. Here is an example to log the latency of a query request:

Response resp = txn.query(query);
Latency latency = resp.getLatency();"parsing latency:" + latency.getParsingNs());"processing latency:" + latency.getProcessingNs());"encoding latency:" + latency.getEncodingNs());

Similarly you can get the latency of a mutation request:

Assigned assignedIds = dgraphClient.newTransaction().mutate(mu);
Latency latency = assignedIds.getLatency();


Building the source

Warning: The gradle build runs integration tests on a locally running Dgraph server. The tests will remove all data from your Dgraph instance. So make sure that you don't have any important data on your Dgraph instance.

./gradlew build

If you have made changes to the task.proto file, this step will also regenerate the source files generated by Protocol Buffer tools.

Code Style

We use google-java-format to format the source code. If you run ./gradlew build, you will be warned if there is code that is not conformant. You can run ./gradlew goJF to format the source code, before committing it.

Running unit tests

Warning: This command will runs integration tests on a locally running Dgraph server. The tests will remove all data from your Dgraph instance. So make sure that you don't have any important data on your Dgraph instance.

Make sure you have a Dgraph server running on localhost before you run this task.

./gradlew test
You can’t perform that action at this time.