导出 NebulaGraph 数据

Exchange 支持将 NebulaGraph 中的数据导出到 CSV 文件或另一个图空间（不同 NebulaGraph 集群也支持）中。本文介绍具体的操作步骤。

Enterpriseonly

仅企业版 Exchange 支持导出 NebulaGraph 数据。

环境准备

本示例在 Linux 系统的虚拟机环境下完成，导出数据前准备的软硬件信息如下。

硬件

类型	信息
CPU	4 Intel(R) Xeon(R) Platinum 8260 CPU @ 2.30GHz
内存	16G
硬盘	50G

系统

CentOS 7.9.2009

软件

名称	版本
JDK	1.8.0
Hadoop	2.10.1
Scala	2.12.11
Spark	2.4.7
NebulaGraph	3.4.3

数据集

在本示例中，作为数据源的 NebulaGraph 存储着 basketballplayer 数据集，其中的 Schema 要素如下表所示。

要素	名称	属性
Tag	player	name string, age int
Tag	team	name string
Edge type	follow	degree int
Edge type	serve	start_year int, end_year int

操作步骤

1. 从 NebulaGraph 企业版套餐 中获取企业版 Exchange 的 JAR 文件。

2. 修改配置文件。

   企业版 Exchange 提供了导出 NebulaGraph 数据专用的配置文件模板export_to_csv.conf和export_to_nebula.conf，其中各配置项的说明参见 Exchange 配置。本示例使用的配置文件核心内容如下：

   • 导出到 CSV 文件：

   # Use the command to submit the exchange job:
   
   # spark-submit \
   # --master "spark://master_ip:7077" \
   # --driver-memory=2G --executor-memory=30G  \
   # --total-executor-cores=60 --executor-cores=20 \
   # --class com.vesoft.nebula.exchange.Exchange \
   # nebula-exchange-3.0-SNAPSHOT.jar -c export_to_csv.conf
   
   {
     # Spark config
     spark: {
       app: {
         name: NebulaGraph Exchange
       }
     }
   
     # Nebula Graph config
     # if you export nebula data to csv, please ignore these nebula config
     nebula: {
       address:{
         graph:["127.0.0.1:9669"]
   
         # the address of any of the meta services.
         # if your NebulaGraph server is in virtual network like k8s, please config the leader address of meta.
         meta:["127.0.0.1:9559"]
       }
       user: root
       pswd: nebula
       space: test
   
       # nebula client connection parameters
       connection {
         # socket connect & execute timeout, unit: millisecond
         timeout: 30000
       }
   
       error: {
         # max number of failures, if the number of failures is bigger than max, then exit the   application.
         max: 32
         # failed data will be recorded in output path, format with ngql
         output: /tmp/errors
       }
   
       # use google's RateLimiter to limit the requests send to NebulaGraph
       rate: {
         # the stable throughput of RateLimiter
         limit: 1024
         # Acquires a permit from RateLimiter, unit: MILLISECONDS
         # if it can't be obtained within the specified timeout, then give up the request.
         timeout: 1000
       }
     }
   
     # Processing tags
     tags: [
       {
         # you can ignore the tag name when export nebula data to csv
         name: tag-name-1
         type: {
           source: nebula
           sink: csv
         }
         metaAddress:"127.0.0.1:9559"
         space:"test"
         label:"person"
         # config the fields you want to export from nebula
         fields: [nebula-field-0, nebula-field-1, nebula-field-2]
         noFields:false  # default false, if true, just export id
         partition: 60
         # config the path to save your csv file. if your file in not in hdfs, config "file:///path/  test.csv"
         path: "hdfs://ip:port/path/person"
         separator: ","
         header: true
       }
     ]
   
     # process edges
     edges: [
       {
         # you can ignore the edge name when export nebula data to csv
         name: edge-name-1
         type: {
           source: nebula
           sink: csv
         }
         metaAddress:"127.0.0.1:9559"
         space:"test"
         label:"friend"
         # config the fields you want to export from nebula
         fields: [nebula-field-0, nebula-field-1, nebula-field-2]
         noFields:false  # default false, if true, just export id
         partition: 60
         # config the path to save your csv file. if your file in not in hdfs, config "file:///path/  test.csv"
         path: "hdfs://ip:port/path/friend"
         separator: ","
         header: true
       }
     ]
   }
   

   • 导出到另一个图空间：

   # Use the command to submit the exchange job:
   
   # spark-submit \
   # --master "spark://master_ip:7077" \
   # --driver-memory=2G --executor-memory=30G  \
   # --total-executor-cores=60 --executor-cores=20 \
   # --class com.vesoft.nebula.exchange.Exchange \
   # nebula-exchange-3.0-SNAPSHOT.jar -c export_to_nebula.conf
   
   {
     # Spark config
     spark: {
       app: {
         name: NebulaGraph Exchange
       }
     }
   
     # Nebula Graph config, just config the sink nebula information
     nebula: {
       address:{
         graph:["127.0.0.1:9669"]
   
         # the address of any of the meta services.
         # if your NebulaGraph server is in virtual network like k8s, please config the leader address of meta.
         meta:["127.0.0.1:9559"]
       }
       user: root
       pswd: nebula
       space: test
   
       # nebula client connection parameters
       connection {
         # socket connect & execute timeout, unit: millisecond
         timeout: 30000
       }
   
       error: {
         # max number of failures, if the number of failures is bigger than max, then exit the   application.
         max: 32
         # failed data will be recorded in output path, format with ngql
         output: /tmp/errors
       }
   
       # use google's RateLimiter to limit the requests send to NebulaGraph
       rate: {
         # the stable throughput of RateLimiter
         limit: 1024
         # Acquires a permit from RateLimiter, unit: MILLISECONDS
         # if it can't be obtained within the specified timeout, then give up the request.
         timeout: 1000
       }
     }
   
     # Processing tags
     tags: [
       {
         name: tag-name-1
         type: {
           source: nebula
           sink: client
         }
         # data source nebula config
         metaAddress:"127.0.0.1:9559"
         space:"test"
         label:"person"
         # mapping the fields of the original NebulaGraph to the fields of the target NebulaGraph.
         fields: [source_nebula-field-0, source_nebula-field-1, source_nebula-field-2]
         nebula.fields: [target_nebula-field-0, target_nebula-field-1, target_nebula-field-2]
         limit:10000
         vertex: _vertexId  # must be `_vertexId`
         batch: 2000
         partition: 60
       }
     ]
   
     # process edges
     edges: [
       {
         name: edge-name-1
         type: {
           source: csv
           sink: client
         }
         # data source nebula config
         metaAddress:"127.0.0.1:9559"
         space:"test"
         label:"friend"
         fields: [source_nebula-field-0, source_nebula-field-1, source_nebula-field-2]
         nebula.fields: [target_nebula-field-0, target_nebula-field-1, target_nebula-field-2]
         limit:1000
         source: _srcId # must be `_srcId`
         target: _dstId # must be `_dstId`
         ranking: source_nebula-field-2
         batch: 2000
         partition: 60
       }
     ]
   }
   

3. 使用如下命令导出 NebulaGraph 中的数据。

   Note

   Driver 和 Executor 进程的相关参数可以根据自身配置灵活修改。

   <spark_install_path>/bin/spark-submit --master "spark://<master_ip>:7077" \
   --driver-memory=2G --executor-memory=30G \
   --total-executor-cores=60 --executor-cores=20 \
   --class com.vesoft.nebula.exchange.Exchange nebula-exchange-x.y.z.jar_path> \
   -c <conf_file_path>
   

   例如导出到 CSV 文件的示例命令如下。

   $ ./spark-submit --master "spark://192.168.10.100:7077" \
   --driver-memory=2G --executor-memory=30G \
   --total-executor-cores=60 --executor-cores=20 \
   --class com.vesoft.nebula.exchange.Exchange ~/exchange-ent/nebula-exchange-ent-3.4.0.jar \
   -c ~/exchange-ent/export_to_csv.conf
   

4. 检查导出的数据。

   • 导出到 CSV 文件：

     查看目标路径下是否成功生成了 CSV 文件，并检查文件内容。

     $ hadoop fs -ls /vertex/player
     Found 11 items
     -rw-r--r--   3 nebula supergroup          0 2021-11-05 07:36 /vertex/player/_SUCCESS
     -rw-r--r--   3 nebula supergroup        160 2021-11-05 07:36 /vertex/player/    part-00000-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        163 2021-11-05 07:36 /vertex/player/    part-00001-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        172 2021-11-05 07:36 /vertex/player/    part-00002-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        172 2021-11-05 07:36 /vertex/player/    part-00003-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        144 2021-11-05 07:36 /vertex/player/    part-00004-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        173 2021-11-05 07:36 /vertex/player/    part-00005-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        160 2021-11-05 07:36 /vertex/player/    part-00006-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        148 2021-11-05 07:36 /vertex/player/    part-00007-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        125 2021-11-05 07:36 /vertex/player/    part-00008-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     -rw-r--r--   3 nebula supergroup        119 2021-11-05 07:36 /vertex/player/    part-00009-17293020-ba2e-4243-b834-34495c0536b3-c000.csv
     

   • 导出到另一个图空间：

     登录新的图空间，通过SUBMIT JOB STATS和SHOW STATS命令查看统计信息，确认是否导出成功。

---

最后更新: September 4, 2023