Flink Jar Jobs Using DEW to Acquire Access Credentials for Reading and Writing Data from and to OBS

Scenario

To write the output data of a Flink Jar job to OBS, AK/SK is required for accessing OBS. To ensure the security of AK/SK data, you can use Data Encryption Workshop (DEW) and Cloud Secret Management Service (CSMS) for unified management of AK/SK, effectively avoiding sensitive information leakage and business risks caused by hard-coded or plaintext configuration of programs.

This section walks you through on how a Flink Jar job acquires an AK/SK to read and write data from and to OBS.

Prerequisites

• A shared secret has been created on the DEW console and the secret value has been stored. For details, see Creating a Shared Secret.
• An agency has been created and authorized for DLI to access DEW. The agency must have been granted the following permissions:
  • Permission of the ShowSecretVersion interface for querying secret versions and secret values in DEW: csms:secretVersion:get.
  • Permission of the ListSecretVersions interface for listing secret versions in DEW: csms:secretVersion:list.
  • Permission to decrypt DEW secrets: kms:dek:decrypt

  For details about agency permission examples, see Customizing DLI Agency Permissions and Agency Permission Policies in Common Scenarios.

• DEW can be used to manage access credentials only in Flink 1.15. When creating a Flink job, select version 1.15 and configure the information of the agency that allows DLI to access DEW for the job. For details about how to create a custom agency and configure it, see Customizing DLI Agency Permissions.
• To use this function, you need to configure AK/SK for all OBS buckets.

Syntax

On the Flink Jar job editing page, set Runtime Configuration as needed. The configuration information is as follows:

Different OBS buckets use different AK/SK authentication information. You can use the following configuration method to specify the AK/SK information based on the bucket. For details about the parameters, see Table 1.

flink.hadoop.fs.obs.bucket.USER_BUCKET_NAME.dew.access.key=USER_AK_CSMS_KEY
flink.hadoop.fs.obs.bucket.USER_BUCKET_NAME.dew.secret.key=USER_SK_CSMS_KEY
flink.hadoop.fs.obs.security.provider=com.dli.provider.UserObsBasicCredentialProvider
flink.hadoop.fs.dew.csms.secretName=CredentialName
flink.hadoop.fs.dew.endpoint=ENDPOINT
flink.hadoop.fs.dew.csms.version=VERSION_ID
flink.hadoop.fs.dew.csms.cache.time.second=CACHE_TIME
flink.dli.job.agency.name=USER_AGENCY_NAME

Parameter Description

Sample Code

This section describes how to write processed DataGen data to OBS. You need to modify the parameters in the sample Java code based on site requirements.

1. Create an agency for DLI to access DEW and complete authorization. For details, see Customizing DLI Agency Permissions.
2. Create a shared secret in DEW. For details, see Creating a Shared Secret.
   1. Log in to the DEW management console.
   2. In the navigation pane on the left, choose Cloud Secret Management Service > Secrets.
   3. On the displayed page, click Create Secret. Set basic secret information.
3. Set job parameters on the DLI Flink Jar job editing page.
   • Class Name

     com.dli.demo.dew.DataGen2FileSystemSink

   • Class Arguments

     --checkpoint.path obs://test/flink/jobs/checkpoint/120891/ 
     --output.path obs://dli/flink.db/79914/DataGen2FileSystemSink

   • Runtime Configuration

     flink.hadoop.fs.obs.bucket.USER_BUCKET_NAME.dew.access.key=USER_AK_CSMS_KEY
     flink.hadoop.fs.obs.bucket.USER_BUCKET_NAME.dew.secret.key=USER_SK_CSMS_KEY
     flink.hadoop.fs.obs.security.provider=com.dli.provider.UserObsBasicCredentialProvider
     flink.hadoop.fs.dew.csms.secretName=obsAksK
     flink.hadoop.fs.dew.endpoint=kmsendpoint
     flink.hadoop.fs.dew.csms.version=v6
     flink.hadoop.fs.dew.csms.cache.time.second=3600
     flink.dli.job.agency.name=***

4. Flink Jar job example.
   • Environment preparation

     Development tools such as IntelliJ IDEA and other development tools, JDK, and Maven have been installed and configured.

     Dependency package in POM file configurations

      <properties>
             <flink.version>1.15.0</flink.version>
         </properties>
     
         <dependencies>
             <dependency>
                 <groupId>org.apache.flink</groupId>
                 <artifactId>flink-statebackend-rocksdb</artifactId>
                 <version>${flink.version}</version>
                 <scope>provided</scope>
             </dependency>
     
             <dependency>
                 <groupId>org.apache.flink</groupId>
                 <artifactId>flink-streaming-java</artifactId>
                 <version>${flink.version}</version>
                 <scope>provided</scope>
             </dependency>
     
             <!-- fastjson -->
             <dependency>
                 <artifactId>fastjson</artifactId>
                 <version>2.0.15</version>
             </dependency>
         </dependencies>

   • Sample code

     package com.huawei.dli.demo.dew;
     
     import org.apache.flink.api.common.serialization.SimpleStringEncoder;
     import org.apache.flink.api.java.utils.ParameterTool;
     import org.apache.flink.contrib.streaming.state.EmbeddedRocksDBStateBackend;
     import org.apache.flink.core.fs.Path;
     import org.apache.flink.streaming.api.datastream.DataStream;
     import org.apache.flink.streaming.api.environment.CheckpointConfig;
     import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
     import org.apache.flink.streaming.api.functions.sink.filesystem.StreamingFileSink;
     import org.apache.flink.streaming.api.functions.sink.filesystem.rollingpolicies.OnCheckpointRollingPolicy;
     import org.apache.flink.streaming.api.functions.source.ParallelSourceFunction;
     import org.slf4j.Logger;
     import org.slf4j.LoggerFactory;
     
     import java.time.LocalDateTime;
     import java.time.ZoneOffset;
     import java.time.format.DateTimeFormatter;
     import java.util.Random;
     
     public class DataGen2FileSystemSink {
         private static final Logger LOG = LoggerFactory.getLogger(DataGen2FileSystemSink.class);
     
         public static void main(String[] args) {
             ParameterTool params = ParameterTool.fromArgs(args);
             LOG.info("Params: " + params.toString());
             try {
                 StreamExecutionEnvironment streamEnv = StreamExecutionEnvironment.getExecutionEnvironment();
     
                 // set checkpoint
                 String checkpointPath = params.get("checkpoint.path", "obs://bucket/checkpoint/jobId_jobName/");
                 LocalDateTime localDateTime = LocalDateTime.ofEpochSecond(System.currentTimeMillis() / 1000,
                     0, ZoneOffset.ofHours(8));
                 String dt = localDateTime.format(DateTimeFormatter.ofPattern("yyyyMMdd_HH:mm:ss"));
                 checkpointPath = checkpointPath + dt;
     
                 streamEnv.setStateBackend(new EmbeddedRocksDBStateBackend());
                 streamEnv.getCheckpointConfig().setCheckpointStorage(checkpointPath);
                 streamEnv.getCheckpointConfig().setExternalizedCheckpointCleanup(
                     CheckpointConfig.ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION);
                 streamEnv.enableCheckpointing(30 * 1000);
     
                 DataStream<String> stream = streamEnv.addSource(new DataGen())
                     .setParallelism(1)
                     .disableChaining();
     
                 String outputPath = params.get("output.path", "obs://bucket/outputPath/jobId_jobName");
     
                 // Sink OBS
                 final StreamingFileSink<String> sinkForRow = StreamingFileSink
                     .forRowFormat(new Path(outputPath), new SimpleStringEncoder<String>("UTF-8"))
                     .withRollingPolicy(OnCheckpointRollingPolicy.build())
                     .build();
     
                 stream.addSink(sinkForRow);
     
                 streamEnv.execute("sinkForRow");
             } catch (Exception e) {
                 LOG.error(e.getMessage(), e);
             }
         }
     }
     
     class DataGen implements ParallelSourceFunction<String> {
     
         private boolean isRunning = true;
     
         private Random random = new Random();
     
         @Override
         public void run(SourceContext<String> ctx) throws Exception {
             while (isRunning) {
                 JSONObject jsonObject = new JSONObject();
                 jsonObject.put("id", random.nextLong());
                 jsonObject.put("name", "Molly" + random.nextInt());
                 jsonObject.put("address", "hangzhou" + random.nextInt());
                 jsonObject.put("birthday", System.currentTimeMillis());
                 jsonObject.put("city", "hangzhou" + random.nextInt());
                 jsonObject.put("number", random.nextInt());
                 ctx.collect(jsonObject.toJSONString());
                 Thread.sleep(1000);
             }
         }
     
         @Override
         public void cancel() {
             isRunning = false;
         }
     }