imbajin commented on code in PR #464: URL: https://github.com/apache/hugegraph-doc/pull/464#discussion_r4000530726
########## content/cn/docs/quickstart/toolchain/hugegraph-seatunnel-connector.md: ########## @@ -0,0 +1,439 @@ +--- +title: "HugeGraph-SeaTunnel Connector Quick Start" +linkTitle: "使用 SeaTunnel Connector 同步数据" +weight: 5 +--- + +<!-- + TODO(apache/hugegraph-doc#464)：本文 SeaTunnel 官方文档链接暂指向 latest（无版本号）页面。 + HugeGraph Source Connector 尚未随正式版本发布，官网 latest 无对应页面， + 故 Source 文档链接暂指向 GitHub dev 分支文件。 + 待 Source 随正式版本发布后，将链接切换为官网版本化地址， + 例如 https://seatunnel.apache.org/docs/2.3.14/connectors/source/HugeGraph/ +--> + +### 1 版本与兼容性 + +| 组件 | 版本要求 | 说明 | +|------|---------|------| +| Java | 11+ | HugeGraph Client 1.5.0+ 运行环境要求 | +| Apache SeaTunnel | 2.3.13+ | Sink 自 2.3.12 合入；2.3.13 起内置 HugeGraph Client 1.5.0 | +| HugeGraph Server | 1.5.0+ | 需与 Connector 内置 Client 版本匹配 | + +> Source Connector 目前只在 SeaTunnel next（dev）分支，正式版还没带。要用就从源码编译，或等下一个 Release；next 分支内置的 HugeGraph Client 已升到 1.7.0，需搭配对应版本的 Server。 + +### 2 概述 + +[Apache SeaTunnel](https://seatunnel.apache.org/) 是开源数据集成平台，批处理、流式同步都支持，自带 100+ 连接器。HugeGraph 的 Connector-V2 已合入：Sink 负责写入（顶点/边的批量写入、更新、删除），Source 负责读取（批量读取、整图迁移）。 + + + +### 3 选型：SeaTunnel 还是 Loader + +可以把 SeaTunnel 理解为 [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) + [Tools](/cn/docs/quickstart/toolchain/hugegraph-tools) 的合集。两者不冲突：Loader 只管 HugeGraph，开箱即用；SeaTunnel 面向所有数据系统。按场景选： + +| 你的场景 | 推荐 | 理由 | +|---------|------|------| +| 一次性 / 定时把本地文件、HDFS、MySQL 等导入 HugeGraph | Loader | 免部署，映射文件即写即用 | +| 数据已在（或必须经过）Flink、Spark、Kafka 等大数据管道 | SeaTunnel | 复用现有管道，不引入第二套导入工具 | +| 实时流式导入、CDC 增量同步 | SeaTunnel | 原生流式作业 + checkpoint 断点恢复 | +| 导入时自动创建 Schema（PropertyKey / VertexLabel / EdgeLabel） | SeaTunnel | Sink 默认 `CREATE_SCHEMA_WHEN_NOT_EXIST` | +| 只维护 HugeGraph 一张图，规模可控、追求简单 | Loader | 工具链内闭环，无额外集群 | + +选 SeaTunnel 要接受两点：部署一套 SeaTunnel 集群（或复用现有的），作业用通用 HOCON 配置，而不是 HugeGraph 映射文件。 + +### 4 快速开始 + +前置条件：HugeGraph Server 1.5.0+（[部署指南](/cn/docs/quickstart/hugegraph/hugegraph-server)）。 + +#### 4.1 部署 SeaTunnel + +**Docker（推荐）**。拉取镜像，把作业配置所在目录挂载进容器提交，宿主机不用装 Java 环境： + +```bash +docker pull apache/seatunnel:2.3.13 + +docker run --rm -it \ + -v /path/to/job:/config \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +更多用法见官方 [Docker 部署文档](https://seatunnel.apache.org/docs/getting-started/docker/docker/)。 + +**Kubernetes**：生产集群用 Helm 部署，见官方 [K8s（Helm）部署文档](https://seatunnel.apache.org/docs/getting-started/kubernetes/helm/)。 + +**二进制包（参考）**：从 [下载页](https://seatunnel.apache.org/download/) 取安装包，解压后装插件、跑脚本。JVM 脚本方式仅作本地调试参考，生产优先 Docker / K8s： + +```bash +sh bin/install-plugin.sh 2.3.13 +sh bin/seatunnel.sh --config ./config/hugegraph-sync.conf -e local +``` + +部署细节以官方 [本地部署文档](https://seatunnel.apache.org/docs/getting-started/locally/deployment/) 为准。 + +#### 4.2 最小示例：CSV 文件 → person 顶点 + +准备一个无表头的 CSV（列顺序与 `schema.fields` 声明顺序一致），内容如下： + +```csv +marko,29 +vadas,27 +josh,32 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/person.csv" + file_format_type = "csv" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + # 以下为可选参数，默认值见第 6 节；认证开启时再填 username/password + # protocol = "http" + # username = "admin" + # password = "admin" + # schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +一行 CSV 到图顶点的映射过程： + +```text +┌───────────────── 输入行 (SeaTunnel Row) ─────────────────┐ +│ name = "marko" age = 29 │ +└──────────────────────────┬───────────────────────────────┘ + │ mappings: + │ type = VERTEX, label = person + │ idStrategy = PRIMARY_KEY, idFields = [name] + ▼ + ┌───────────────────────────────┐ + │ HugeGraph 顶点 │ + │ id = person:marko │ + │ label = person │ + │ props = { age: 29 } │ + └───────────────────────────────┘ +``` + +#### 4.3 运行与验证 + +```bash +# 配置和 CSV 分别挂载进容器；路径换成你机器上的实际目录 +docker run --rm -it \ + -v /path/to/job:/config \ + -v /path/to/data:/data \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +运行后，在 Hubble 或 REST API 里执行这条 Gremlin 验证： + +```groovy +g.V().hasLabel('person').valueMap() +``` + +> **Schema 说明**：`mappings` 模式下默认 `schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST`，写入前自动补建缺失的 PropertyKey / VertexLabel / EdgeLabel。要严格管控就自己先建 Schema，并设 `ERROR_WHEN_SCHEMA_NOT_EXIST`。 + +#### 4.4 写入边 + +关系行到边的映射同样通过 `mappings` 完成，用 `sourceConfig` / `targetConfig` 还原边的两个端点。 + +<details> +<summary>展开查看：关系 CSV → knows 边 完整配置</summary> + +CSV（无表头）： + +```csv +marko,vadas,2020 +marko,josh,2021 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/knows.csv" + file_format_type = "csv" + schema = { + fields = { + person1_name = "string" + person2_name = "string" + since = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "EDGE" + label = "knows" + sourceConfig = { + label = "person" + idFields = ["person1_name"] + } + targetConfig = { + label = "person" + idFields = ["person2_name"] + } + properties = ["since"] + fieldMapping = { + person1_name = "name" + person2_name = "name" + } + } + ] + } +} +``` + +</details> + +```text +person1_name = "marko" person2_name = "vadas" since = 2020 + │ │ + ▼ ▼ + sourceConfig targetConfig + label = person label = person + idFields = [person1_name] idFields = [person2_name] + └───────────┬────────────┘ + ▼ + edge: person:marko -[knows]-> person:vadas + properties = { since: 2020 } +``` + +> 边的端点 ID 策略从服务端已有的 VertexLabel 读取，`sourceConfig.idFields` / `targetConfig.idFields` 必须能拼出端点 ID。默认 `check_vertex = false`，顶点和边允许乱序写入，跑完图最终一致；设成 `true` 则服务端直接拒绝端点不存在的边。 + +### 5 常见场景 + +#### 5.1 Kafka 实时流导入 + +```mermaid +flowchart LR + P["业务系统"] --> K[("Kafka Topic")] + K -->|"STREAMING + checkpoint"| ST["SeaTunnel<br/>Kafka Source → HugeGraph Sink"] + ST --> HG[("HugeGraph Server")] +``` + +<details> +<summary>展开查看：Kafka → HugeGraph 流式作业配置</summary> + +```hocon +env { + job.mode = "STREAMING" + checkpoint.interval = 10000 +} + +source { + Kafka { + bootstrap.servers = "localhost:9092" + topics = "user-events" Review Comment: Fixed: the Kafka job uses topic="user-events", matching the current upstream option. The complete document configuration was parsed during validation. Updated in 1cde03e0fba7cc5a3e76d7a326759eefa874467f. ########## content/cn/docs/quickstart/toolchain/hugegraph-seatunnel-connector.md: ########## @@ -0,0 +1,439 @@ +--- +title: "HugeGraph-SeaTunnel Connector Quick Start" +linkTitle: "使用 SeaTunnel Connector 同步数据" +weight: 5 +--- + +<!-- + TODO(apache/hugegraph-doc#464)：本文 SeaTunnel 官方文档链接暂指向 latest（无版本号）页面。 + HugeGraph Source Connector 尚未随正式版本发布，官网 latest 无对应页面， + 故 Source 文档链接暂指向 GitHub dev 分支文件。 + 待 Source 随正式版本发布后，将链接切换为官网版本化地址， + 例如 https://seatunnel.apache.org/docs/2.3.14/connectors/source/HugeGraph/ +--> + +### 1 版本与兼容性 + +| 组件 | 版本要求 | 说明 | +|------|---------|------| +| Java | 11+ | HugeGraph Client 1.5.0+ 运行环境要求 | +| Apache SeaTunnel | 2.3.13+ | Sink 自 2.3.12 合入；2.3.13 起内置 HugeGraph Client 1.5.0 | +| HugeGraph Server | 1.5.0+ | 需与 Connector 内置 Client 版本匹配 | + +> Source Connector 目前只在 SeaTunnel next（dev）分支，正式版还没带。要用就从源码编译，或等下一个 Release；next 分支内置的 HugeGraph Client 已升到 1.7.0，需搭配对应版本的 Server。 + +### 2 概述 + +[Apache SeaTunnel](https://seatunnel.apache.org/) 是开源数据集成平台，批处理、流式同步都支持，自带 100+ 连接器。HugeGraph 的 Connector-V2 已合入：Sink 负责写入（顶点/边的批量写入、更新、删除），Source 负责读取（批量读取、整图迁移）。 + + + +### 3 选型：SeaTunnel 还是 Loader + +可以把 SeaTunnel 理解为 [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) + [Tools](/cn/docs/quickstart/toolchain/hugegraph-tools) 的合集。两者不冲突：Loader 只管 HugeGraph，开箱即用；SeaTunnel 面向所有数据系统。按场景选： + +| 你的场景 | 推荐 | 理由 | +|---------|------|------| +| 一次性 / 定时把本地文件、HDFS、MySQL 等导入 HugeGraph | Loader | 免部署，映射文件即写即用 | +| 数据已在（或必须经过）Flink、Spark、Kafka 等大数据管道 | SeaTunnel | 复用现有管道，不引入第二套导入工具 | +| 实时流式导入、CDC 增量同步 | SeaTunnel | 原生流式作业 + checkpoint 断点恢复 | +| 导入时自动创建 Schema（PropertyKey / VertexLabel / EdgeLabel） | SeaTunnel | Sink 默认 `CREATE_SCHEMA_WHEN_NOT_EXIST` | +| 只维护 HugeGraph 一张图，规模可控、追求简单 | Loader | 工具链内闭环，无额外集群 | + +选 SeaTunnel 要接受两点：部署一套 SeaTunnel 集群（或复用现有的），作业用通用 HOCON 配置，而不是 HugeGraph 映射文件。 + +### 4 快速开始 + +前置条件：HugeGraph Server 1.5.0+（[部署指南](/cn/docs/quickstart/hugegraph/hugegraph-server)）。 + +#### 4.1 部署 SeaTunnel + +**Docker（推荐）**。拉取镜像，把作业配置所在目录挂载进容器提交，宿主机不用装 Java 环境： + +```bash +docker pull apache/seatunnel:2.3.13 + +docker run --rm -it \ + -v /path/to/job:/config \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +更多用法见官方 [Docker 部署文档](https://seatunnel.apache.org/docs/getting-started/docker/docker/)。 + +**Kubernetes**：生产集群用 Helm 部署，见官方 [K8s（Helm）部署文档](https://seatunnel.apache.org/docs/getting-started/kubernetes/helm/)。 + +**二进制包（参考）**：从 [下载页](https://seatunnel.apache.org/download/) 取安装包，解压后装插件、跑脚本。JVM 脚本方式仅作本地调试参考，生产优先 Docker / K8s： + +```bash +sh bin/install-plugin.sh 2.3.13 +sh bin/seatunnel.sh --config ./config/hugegraph-sync.conf -e local +``` + +部署细节以官方 [本地部署文档](https://seatunnel.apache.org/docs/getting-started/locally/deployment/) 为准。 + +#### 4.2 最小示例：CSV 文件 → person 顶点 + +准备一个无表头的 CSV（列顺序与 `schema.fields` 声明顺序一致），内容如下： + +```csv +marko,29 +vadas,27 +josh,32 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/person.csv" + file_format_type = "csv" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + # 以下为可选参数，默认值见第 6 节；认证开启时再填 username/password + # protocol = "http" + # username = "admin" + # password = "admin" + # schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +一行 CSV 到图顶点的映射过程： + +```text +┌───────────────── 输入行 (SeaTunnel Row) ─────────────────┐ +│ name = "marko" age = 29 │ +└──────────────────────────┬───────────────────────────────┘ + │ mappings: + │ type = VERTEX, label = person + │ idStrategy = PRIMARY_KEY, idFields = [name] + ▼ + ┌───────────────────────────────┐ + │ HugeGraph 顶点 │ + │ id = person:marko │ + │ label = person │ + │ props = { age: 29 } │ + └───────────────────────────────┘ +``` + +#### 4.3 运行与验证 + +```bash +# 配置和 CSV 分别挂载进容器；路径换成你机器上的实际目录 +docker run --rm -it \ + -v /path/to/job:/config \ + -v /path/to/data:/data \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +运行后，在 Hubble 或 REST API 里执行这条 Gremlin 验证： + +```groovy +g.V().hasLabel('person').valueMap() +``` + +> **Schema 说明**：`mappings` 模式下默认 `schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST`，写入前自动补建缺失的 PropertyKey / VertexLabel / EdgeLabel。要严格管控就自己先建 Schema，并设 `ERROR_WHEN_SCHEMA_NOT_EXIST`。 + +#### 4.4 写入边 + +关系行到边的映射同样通过 `mappings` 完成，用 `sourceConfig` / `targetConfig` 还原边的两个端点。 + +<details> +<summary>展开查看：关系 CSV → knows 边 完整配置</summary> + +CSV（无表头）： + +```csv +marko,vadas,2020 +marko,josh,2021 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/knows.csv" + file_format_type = "csv" + schema = { + fields = { + person1_name = "string" + person2_name = "string" + since = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "EDGE" + label = "knows" + sourceConfig = { + label = "person" + idFields = ["person1_name"] + } + targetConfig = { + label = "person" + idFields = ["person2_name"] + } + properties = ["since"] + fieldMapping = { + person1_name = "name" + person2_name = "name" + } + } + ] + } +} +``` + +</details> + +```text +person1_name = "marko" person2_name = "vadas" since = 2020 + │ │ + ▼ ▼ + sourceConfig targetConfig + label = person label = person + idFields = [person1_name] idFields = [person2_name] + └───────────┬────────────┘ + ▼ + edge: person:marko -[knows]-> person:vadas + properties = { since: 2020 } +``` + +> 边的端点 ID 策略从服务端已有的 VertexLabel 读取，`sourceConfig.idFields` / `targetConfig.idFields` 必须能拼出端点 ID。默认 `check_vertex = false`，顶点和边允许乱序写入，跑完图最终一致；设成 `true` 则服务端直接拒绝端点不存在的边。 + +### 5 常见场景 + +#### 5.1 Kafka 实时流导入 + +```mermaid +flowchart LR + P["业务系统"] --> K[("Kafka Topic")] + K -->|"STREAMING + checkpoint"| ST["SeaTunnel<br/>Kafka Source → HugeGraph Sink"] + ST --> HG[("HugeGraph Server")] +``` + +<details> +<summary>展开查看：Kafka → HugeGraph 流式作业配置</summary> + +```hocon +env { + job.mode = "STREAMING" + checkpoint.interval = 10000 +} + +source { + Kafka { + bootstrap.servers = "localhost:9092" + topics = "user-events" + consumer.group = "hugegraph-import" + format = json + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +</details> + +> 要点：`job.mode = "STREAMING"` 加 `checkpoint.interval`，任务就能断点恢复。Sink 是 at-least-once 语义，`PRIMARY_KEY` / `CUSTOMIZE_*` 这类可还原的 ID 重放只是幂等更新。Kafka 的 offset、分区、format 等参数见 [Kafka Source 官方文档](https://seatunnel.apache.org/docs/connectors/source/Kafka/)。 + +#### 5.2 在 Flink / Spark 上运行 + +作业默认跑在 SeaTunnel 自带的 Zeta 引擎上。已有 Flink / Spark 集群的话，同一份配置换对应的 starter 脚本提交即可，内容不用改。脚本用法和版本支持见官方 [Flink 引擎文档](https://seatunnel.apache.org/docs/engines/flink/) 与 [Spark 引擎文档](https://seatunnel.apache.org/docs/engines/spark/)。 + +#### 5.3 HugeGraph → HugeGraph 图迁移（Source Connector，next 分支） + +```mermaid +flowchart LR + A[("HugeGraph 源实例")] -->|"Source 批量读取"| ST["SeaTunnel"] + ST -->|"Sink 批量写入"| B[("HugeGraph 目标实例")] +``` + +<details> +<summary>展开查看：单图迁移（person 顶点）</summary> + +```hocon +env { + job.mode = "BATCH" +} + +source { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + label = "person" + label_type = "VERTEX" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.2" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +</details> + +<details> +<summary>展开查看：批量迁移（A 实例 3 个图 → B 实例，图名不变）</summary> + +一次作业迁一张图；省略 `label` 时读取该图全部顶点 label，边同理（`label_type = "EDGE"`），顶点和边分开跑。多张图用变量替换 + 循环逐个执行： + +```hocon +env { + job.mode = "BATCH" +} + +source { + HugeGraph { + host = "graph-a:8080" Review Comment: Fixed: source-hugegraph and target-hugegraph contain no port; port=8080 is separate. This rule is also explained in the setup section. Updated in 1cde03e0fba7cc5a3e76d7a326759eefa874467f. ########## content/cn/docs/quickstart/toolchain/hugegraph-seatunnel-connector.md: ########## @@ -0,0 +1,439 @@ +--- +title: "HugeGraph-SeaTunnel Connector Quick Start" +linkTitle: "使用 SeaTunnel Connector 同步数据" +weight: 5 +--- + +<!-- + TODO(apache/hugegraph-doc#464)：本文 SeaTunnel 官方文档链接暂指向 latest（无版本号）页面。 + HugeGraph Source Connector 尚未随正式版本发布，官网 latest 无对应页面， + 故 Source 文档链接暂指向 GitHub dev 分支文件。 + 待 Source 随正式版本发布后，将链接切换为官网版本化地址， + 例如 https://seatunnel.apache.org/docs/2.3.14/connectors/source/HugeGraph/ +--> + +### 1 版本与兼容性 + +| 组件 | 版本要求 | 说明 | +|------|---------|------| +| Java | 11+ | HugeGraph Client 1.5.0+ 运行环境要求 | +| Apache SeaTunnel | 2.3.13+ | Sink 自 2.3.12 合入；2.3.13 起内置 HugeGraph Client 1.5.0 | +| HugeGraph Server | 1.5.0+ | 需与 Connector 内置 Client 版本匹配 | + +> Source Connector 目前只在 SeaTunnel next（dev）分支，正式版还没带。要用就从源码编译，或等下一个 Release；next 分支内置的 HugeGraph Client 已升到 1.7.0，需搭配对应版本的 Server。 + +### 2 概述 + +[Apache SeaTunnel](https://seatunnel.apache.org/) 是开源数据集成平台，批处理、流式同步都支持，自带 100+ 连接器。HugeGraph 的 Connector-V2 已合入：Sink 负责写入（顶点/边的批量写入、更新、删除），Source 负责读取（批量读取、整图迁移）。 + + + +### 3 选型：SeaTunnel 还是 Loader + +可以把 SeaTunnel 理解为 [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) + [Tools](/cn/docs/quickstart/toolchain/hugegraph-tools) 的合集。两者不冲突：Loader 只管 HugeGraph，开箱即用；SeaTunnel 面向所有数据系统。按场景选： + +| 你的场景 | 推荐 | 理由 | +|---------|------|------| +| 一次性 / 定时把本地文件、HDFS、MySQL 等导入 HugeGraph | Loader | 免部署，映射文件即写即用 | +| 数据已在（或必须经过）Flink、Spark、Kafka 等大数据管道 | SeaTunnel | 复用现有管道，不引入第二套导入工具 | +| 实时流式导入、CDC 增量同步 | SeaTunnel | 原生流式作业 + checkpoint 断点恢复 | +| 导入时自动创建 Schema（PropertyKey / VertexLabel / EdgeLabel） | SeaTunnel | Sink 默认 `CREATE_SCHEMA_WHEN_NOT_EXIST` | +| 只维护 HugeGraph 一张图，规模可控、追求简单 | Loader | 工具链内闭环，无额外集群 | + +选 SeaTunnel 要接受两点：部署一套 SeaTunnel 集群（或复用现有的），作业用通用 HOCON 配置，而不是 HugeGraph 映射文件。 + +### 4 快速开始 + +前置条件：HugeGraph Server 1.5.0+（[部署指南](/cn/docs/quickstart/hugegraph/hugegraph-server)）。 + +#### 4.1 部署 SeaTunnel + +**Docker（推荐）**。拉取镜像，把作业配置所在目录挂载进容器提交，宿主机不用装 Java 环境： + +```bash +docker pull apache/seatunnel:2.3.13 + +docker run --rm -it \ + -v /path/to/job:/config \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +更多用法见官方 [Docker 部署文档](https://seatunnel.apache.org/docs/getting-started/docker/docker/)。 + +**Kubernetes**：生产集群用 Helm 部署，见官方 [K8s（Helm）部署文档](https://seatunnel.apache.org/docs/getting-started/kubernetes/helm/)。 + +**二进制包（参考）**：从 [下载页](https://seatunnel.apache.org/download/) 取安装包，解压后装插件、跑脚本。JVM 脚本方式仅作本地调试参考，生产优先 Docker / K8s： + +```bash +sh bin/install-plugin.sh 2.3.13 +sh bin/seatunnel.sh --config ./config/hugegraph-sync.conf -e local +``` + +部署细节以官方 [本地部署文档](https://seatunnel.apache.org/docs/getting-started/locally/deployment/) 为准。 + +#### 4.2 最小示例：CSV 文件 → person 顶点 + +准备一个无表头的 CSV（列顺序与 `schema.fields` 声明顺序一致），内容如下： + +```csv +marko,29 +vadas,27 +josh,32 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/person.csv" + file_format_type = "csv" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + # 以下为可选参数，默认值见第 6 节；认证开启时再填 username/password + # protocol = "http" + # username = "admin" + # password = "admin" + # schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +一行 CSV 到图顶点的映射过程： + +```text +┌───────────────── 输入行 (SeaTunnel Row) ─────────────────┐ +│ name = "marko" age = 29 │ +└──────────────────────────┬───────────────────────────────┘ + │ mappings: + │ type = VERTEX, label = person + │ idStrategy = PRIMARY_KEY, idFields = [name] + ▼ + ┌───────────────────────────────┐ + │ HugeGraph 顶点 │ + │ id = person:marko │ + │ label = person │ + │ props = { age: 29 } │ + └───────────────────────────────┘ +``` + +#### 4.3 运行与验证 + +```bash +# 配置和 CSV 分别挂载进容器；路径换成你机器上的实际目录 +docker run --rm -it \ + -v /path/to/job:/config \ + -v /path/to/data:/data \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +运行后，在 Hubble 或 REST API 里执行这条 Gremlin 验证： + +```groovy +g.V().hasLabel('person').valueMap() +``` + +> **Schema 说明**：`mappings` 模式下默认 `schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST`，写入前自动补建缺失的 PropertyKey / VertexLabel / EdgeLabel。要严格管控就自己先建 Schema，并设 `ERROR_WHEN_SCHEMA_NOT_EXIST`。 + +#### 4.4 写入边 + +关系行到边的映射同样通过 `mappings` 完成，用 `sourceConfig` / `targetConfig` 还原边的两个端点。 + +<details> +<summary>展开查看：关系 CSV → knows 边 完整配置</summary> + +CSV（无表头）： + +```csv +marko,vadas,2020 +marko,josh,2021 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/knows.csv" + file_format_type = "csv" + schema = { + fields = { + person1_name = "string" + person2_name = "string" + since = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "EDGE" + label = "knows" + sourceConfig = { + label = "person" + idFields = ["person1_name"] + } + targetConfig = { + label = "person" + idFields = ["person2_name"] + } + properties = ["since"] + fieldMapping = { + person1_name = "name" + person2_name = "name" + } + } + ] + } +} +``` + +</details> + +```text +person1_name = "marko" person2_name = "vadas" since = 2020 + │ │ + ▼ ▼ + sourceConfig targetConfig + label = person label = person + idFields = [person1_name] idFields = [person2_name] + └───────────┬────────────┘ + ▼ + edge: person:marko -[knows]-> person:vadas + properties = { since: 2020 } +``` + +> 边的端点 ID 策略从服务端已有的 VertexLabel 读取，`sourceConfig.idFields` / `targetConfig.idFields` 必须能拼出端点 ID。默认 `check_vertex = false`，顶点和边允许乱序写入，跑完图最终一致；设成 `true` 则服务端直接拒绝端点不存在的边。 + +### 5 常见场景 + +#### 5.1 Kafka 实时流导入 + +```mermaid +flowchart LR + P["业务系统"] --> K[("Kafka Topic")] + K -->|"STREAMING + checkpoint"| ST["SeaTunnel<br/>Kafka Source → HugeGraph Sink"] + ST --> HG[("HugeGraph Server")] +``` + +<details> +<summary>展开查看：Kafka → HugeGraph 流式作业配置</summary> + +```hocon +env { + job.mode = "STREAMING" + checkpoint.interval = 10000 +} + +source { + Kafka { + bootstrap.servers = "localhost:9092" + topics = "user-events" + consumer.group = "hugegraph-import" + format = json + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +</details> + +> 要点：`job.mode = "STREAMING"` 加 `checkpoint.interval`，任务就能断点恢复。Sink 是 at-least-once 语义，`PRIMARY_KEY` / `CUSTOMIZE_*` 这类可还原的 ID 重放只是幂等更新。Kafka 的 offset、分区、format 等参数见 [Kafka Source 官方文档](https://seatunnel.apache.org/docs/connectors/source/Kafka/)。 + +#### 5.2 在 Flink / Spark 上运行 + +作业默认跑在 SeaTunnel 自带的 Zeta 引擎上。已有 Flink / Spark 集群的话，同一份配置换对应的 starter 脚本提交即可，内容不用改。脚本用法和版本支持见官方 [Flink 引擎文档](https://seatunnel.apache.org/docs/engines/flink/) 与 [Spark 引擎文档](https://seatunnel.apache.org/docs/engines/spark/)。 + +#### 5.3 HugeGraph → HugeGraph 图迁移（Source Connector，next 分支） + +```mermaid +flowchart LR + A[("HugeGraph 源实例")] -->|"Source 批量读取"| ST["SeaTunnel"] + ST -->|"Sink 批量写入"| B[("HugeGraph 目标实例")] +``` + +<details> +<summary>展开查看：单图迁移（person 顶点）</summary> + +```hocon +env { + job.mode = "BATCH" +} + +source { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + label = "person" + label_type = "VERTEX" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.2" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +</details> + +<details> +<summary>展开查看：批量迁移（A 实例 3 个图 → B 实例，图名不变）</summary> + +一次作业迁一张图；省略 `label` 时读取该图全部顶点 label，边同理（`label_type = "EDGE"`），顶点和边分开跑。多张图用变量替换 + 循环逐个执行： + +```hocon +env { + job.mode = "BATCH" +} + +source { + HugeGraph { + host = "graph-a:8080" + port = 8080 + graph_name = "${graph}" + graph_space = "DEFAULT" + label_type = "VERTEX" + # 不写 label：读取该图全部顶点 label；迁边时改成 "EDGE" + } +} + +sink { + HugeGraph { + host = "graph-b:8080" + port = 8080 + graph_name = "${graph}" + graph_space = "DEFAULT" + # mappings 按目标图的 label 逐一列出，源、目标同名 label 一一对应 + mappings = [...] Review Comment: Fixed: the runnable examples read one explicit label per job and contain complete mappings. The all-labels note requires source_table bindings and says the single-label examples cannot be reused unchanged. Updated in 1cde03e0fba7cc5a3e76d7a326759eefa874467f. ########## content/cn/docs/quickstart/toolchain/hugegraph-seatunnel-connector.md: ########## @@ -0,0 +1,439 @@ +--- +title: "HugeGraph-SeaTunnel Connector Quick Start" +linkTitle: "使用 SeaTunnel Connector 同步数据" +weight: 5 +--- + +<!-- + TODO(apache/hugegraph-doc#464)：本文 SeaTunnel 官方文档链接暂指向 latest（无版本号）页面。 + HugeGraph Source Connector 尚未随正式版本发布，官网 latest 无对应页面， + 故 Source 文档链接暂指向 GitHub dev 分支文件。 + 待 Source 随正式版本发布后，将链接切换为官网版本化地址， + 例如 https://seatunnel.apache.org/docs/2.3.14/connectors/source/HugeGraph/ +--> + +### 1 版本与兼容性 + +| 组件 | 版本要求 | 说明 | +|------|---------|------| +| Java | 11+ | HugeGraph Client 1.5.0+ 运行环境要求 | +| Apache SeaTunnel | 2.3.13+ | Sink 自 2.3.12 合入；2.3.13 起内置 HugeGraph Client 1.5.0 | +| HugeGraph Server | 1.5.0+ | 需与 Connector 内置 Client 版本匹配 | + +> Source Connector 目前只在 SeaTunnel next（dev）分支，正式版还没带。要用就从源码编译，或等下一个 Release；next 分支内置的 HugeGraph Client 已升到 1.7.0，需搭配对应版本的 Server。 + +### 2 概述 + +[Apache SeaTunnel](https://seatunnel.apache.org/) 是开源数据集成平台，批处理、流式同步都支持，自带 100+ 连接器。HugeGraph 的 Connector-V2 已合入：Sink 负责写入（顶点/边的批量写入、更新、删除），Source 负责读取（批量读取、整图迁移）。 + + + +### 3 选型：SeaTunnel 还是 Loader + +可以把 SeaTunnel 理解为 [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) + [Tools](/cn/docs/quickstart/toolchain/hugegraph-tools) 的合集。两者不冲突：Loader 只管 HugeGraph，开箱即用；SeaTunnel 面向所有数据系统。按场景选： + +| 你的场景 | 推荐 | 理由 | +|---------|------|------| +| 一次性 / 定时把本地文件、HDFS、MySQL 等导入 HugeGraph | Loader | 免部署，映射文件即写即用 | +| 数据已在（或必须经过）Flink、Spark、Kafka 等大数据管道 | SeaTunnel | 复用现有管道，不引入第二套导入工具 | +| 实时流式导入、CDC 增量同步 | SeaTunnel | 原生流式作业 + checkpoint 断点恢复 | +| 导入时自动创建 Schema（PropertyKey / VertexLabel / EdgeLabel） | SeaTunnel | Sink 默认 `CREATE_SCHEMA_WHEN_NOT_EXIST` | +| 只维护 HugeGraph 一张图，规模可控、追求简单 | Loader | 工具链内闭环，无额外集群 | + +选 SeaTunnel 要接受两点：部署一套 SeaTunnel 集群（或复用现有的），作业用通用 HOCON 配置，而不是 HugeGraph 映射文件。 + +### 4 快速开始 + +前置条件：HugeGraph Server 1.5.0+（[部署指南](/cn/docs/quickstart/hugegraph/hugegraph-server)）。 + +#### 4.1 部署 SeaTunnel + +**Docker（推荐）**。拉取镜像，把作业配置所在目录挂载进容器提交，宿主机不用装 Java 环境： + +```bash +docker pull apache/seatunnel:2.3.13 + +docker run --rm -it \ + -v /path/to/job:/config \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +更多用法见官方 [Docker 部署文档](https://seatunnel.apache.org/docs/getting-started/docker/docker/)。 + +**Kubernetes**：生产集群用 Helm 部署，见官方 [K8s（Helm）部署文档](https://seatunnel.apache.org/docs/getting-started/kubernetes/helm/)。 + +**二进制包（参考）**：从 [下载页](https://seatunnel.apache.org/download/) 取安装包，解压后装插件、跑脚本。JVM 脚本方式仅作本地调试参考，生产优先 Docker / K8s： + +```bash +sh bin/install-plugin.sh 2.3.13 +sh bin/seatunnel.sh --config ./config/hugegraph-sync.conf -e local +``` + +部署细节以官方 [本地部署文档](https://seatunnel.apache.org/docs/getting-started/locally/deployment/) 为准。 + +#### 4.2 最小示例：CSV 文件 → person 顶点 + +准备一个无表头的 CSV（列顺序与 `schema.fields` 声明顺序一致），内容如下： + +```csv +marko,29 +vadas,27 +josh,32 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/person.csv" + file_format_type = "csv" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + # 以下为可选参数，默认值见第 6 节；认证开启时再填 username/password + # protocol = "http" + # username = "admin" + # password = "admin" + # schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +一行 CSV 到图顶点的映射过程： + +```text +┌───────────────── 输入行 (SeaTunnel Row) ─────────────────┐ +│ name = "marko" age = 29 │ +└──────────────────────────┬───────────────────────────────┘ + │ mappings: + │ type = VERTEX, label = person + │ idStrategy = PRIMARY_KEY, idFields = [name] + ▼ + ┌───────────────────────────────┐ + │ HugeGraph 顶点 │ + │ id = person:marko │ + │ label = person │ + │ props = { age: 29 } │ + └───────────────────────────────┘ +``` + +#### 4.3 运行与验证 + +```bash +# 配置和 CSV 分别挂载进容器；路径换成你机器上的实际目录 +docker run --rm -it \ + -v /path/to/job:/config \ + -v /path/to/data:/data \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +运行后，在 Hubble 或 REST API 里执行这条 Gremlin 验证： + +```groovy +g.V().hasLabel('person').valueMap() +``` + +> **Schema 说明**：`mappings` 模式下默认 `schema_save_mode = CREATE_SCHEMA_WHEN_NOT_EXIST`，写入前自动补建缺失的 PropertyKey / VertexLabel / EdgeLabel。要严格管控就自己先建 Schema，并设 `ERROR_WHEN_SCHEMA_NOT_EXIST`。 + +#### 4.4 写入边 + +关系行到边的映射同样通过 `mappings` 完成，用 `sourceConfig` / `targetConfig` 还原边的两个端点。 + +<details> +<summary>展开查看：关系 CSV → knows 边 完整配置</summary> + +CSV（无表头）： + +```csv +marko,vadas,2020 +marko,josh,2021 +``` + +```hocon +env { + job.mode = "BATCH" +} + +source { + LocalFile { + path = "/data/knows.csv" + file_format_type = "csv" + schema = { + fields = { + person1_name = "string" + person2_name = "string" + since = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "EDGE" + label = "knows" + sourceConfig = { + label = "person" + idFields = ["person1_name"] + } + targetConfig = { + label = "person" + idFields = ["person2_name"] + } + properties = ["since"] + fieldMapping = { + person1_name = "name" + person2_name = "name" + } + } + ] + } +} +``` + +</details> + +```text +person1_name = "marko" person2_name = "vadas" since = 2020 + │ │ + ▼ ▼ + sourceConfig targetConfig + label = person label = person + idFields = [person1_name] idFields = [person2_name] + └───────────┬────────────┘ + ▼ + edge: person:marko -[knows]-> person:vadas + properties = { since: 2020 } +``` + +> 边的端点 ID 策略从服务端已有的 VertexLabel 读取，`sourceConfig.idFields` / `targetConfig.idFields` 必须能拼出端点 ID。默认 `check_vertex = false`，顶点和边允许乱序写入，跑完图最终一致；设成 `true` 则服务端直接拒绝端点不存在的边。 + +### 5 常见场景 + +#### 5.1 Kafka 实时流导入 + +```mermaid +flowchart LR + P["业务系统"] --> K[("Kafka Topic")] + K -->|"STREAMING + checkpoint"| ST["SeaTunnel<br/>Kafka Source → HugeGraph Sink"] + ST --> HG[("HugeGraph Server")] +``` + +<details> +<summary>展开查看：Kafka → HugeGraph 流式作业配置</summary> + +```hocon +env { + job.mode = "STREAMING" + checkpoint.interval = 10000 +} + +source { + Kafka { + bootstrap.servers = "localhost:9092" + topics = "user-events" + consumer.group = "hugegraph-import" + format = json + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +</details> + +> 要点：`job.mode = "STREAMING"` 加 `checkpoint.interval`，任务就能断点恢复。Sink 是 at-least-once 语义，`PRIMARY_KEY` / `CUSTOMIZE_*` 这类可还原的 ID 重放只是幂等更新。Kafka 的 offset、分区、format 等参数见 [Kafka Source 官方文档](https://seatunnel.apache.org/docs/connectors/source/Kafka/)。 + +#### 5.2 在 Flink / Spark 上运行 + +作业默认跑在 SeaTunnel 自带的 Zeta 引擎上。已有 Flink / Spark 集群的话，同一份配置换对应的 starter 脚本提交即可，内容不用改。脚本用法和版本支持见官方 [Flink 引擎文档](https://seatunnel.apache.org/docs/engines/flink/) 与 [Spark 引擎文档](https://seatunnel.apache.org/docs/engines/spark/)。 + +#### 5.3 HugeGraph → HugeGraph 图迁移（Source Connector，next 分支） + +```mermaid +flowchart LR + A[("HugeGraph 源实例")] -->|"Source 批量读取"| ST["SeaTunnel"] + ST -->|"Sink 批量写入"| B[("HugeGraph 目标实例")] +``` + +<details> +<summary>展开查看：单图迁移（person 顶点）</summary> + +```hocon +env { + job.mode = "BATCH" +} + +source { + HugeGraph { + host = "127.0.0.1" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + label = "person" + label_type = "VERTEX" + schema = { + fields = { + name = "string" + age = "int" + } + } + } +} + +sink { + HugeGraph { + host = "127.0.0.2" + port = 8080 + graph_name = "hugegraph" + graph_space = "DEFAULT" + mappings = [ + { + type = "VERTEX" + label = "person" + idStrategy = "PRIMARY_KEY" + idFields = ["name"] + properties = ["name", "age"] + } + ] + } +} +``` + +</details> + +<details> +<summary>展开查看：批量迁移（A 实例 3 个图 → B 实例，图名不变）</summary> + +一次作业迁一张图；省略 `label` 时读取该图全部顶点 label，边同理（`label_type = "EDGE"`），顶点和边分开跑。多张图用变量替换 + 循环逐个执行： + +```hocon +env { + job.mode = "BATCH" +} + +source { + HugeGraph { + host = "graph-a:8080" + port = 8080 + graph_name = "${graph}" + graph_space = "DEFAULT" + label_type = "VERTEX" + # 不写 label：读取该图全部顶点 label；迁边时改成 "EDGE" + } +} + +sink { + HugeGraph { + host = "graph-b:8080" + port = 8080 + graph_name = "${graph}" + graph_space = "DEFAULT" + # mappings 按目标图的 label 逐一列出，源、目标同名 label 一一对应 + mappings = [...] + } +} +``` + +```bash +# 每个图跑两次：先顶点（VERTEX）、再边（EDGE），图名不变 +for g in graph1 graph2 graph3; do + docker run --rm -it -v /path/to/job:/config apache/seatunnel:2.3.13 \ Review Comment: Fixed: migration is scoped to the pinned 3.0 dev build, with separate VERTEX and EDGE configurations and commands, in that order. The 2.3.13 Docker loop has been removed. Updated in 1cde03e0fba7cc5a3e76d7a326759eefa874467f. ########## content/cn/docs/quickstart/toolchain/hugegraph-seatunnel-connector.md: ########## @@ -0,0 +1,439 @@ +--- +title: "HugeGraph-SeaTunnel Connector Quick Start" +linkTitle: "使用 SeaTunnel Connector 同步数据" +weight: 5 +--- + +<!-- + TODO(apache/hugegraph-doc#464)：本文 SeaTunnel 官方文档链接暂指向 latest（无版本号）页面。 + HugeGraph Source Connector 尚未随正式版本发布，官网 latest 无对应页面， + 故 Source 文档链接暂指向 GitHub dev 分支文件。 + 待 Source 随正式版本发布后，将链接切换为官网版本化地址， + 例如 https://seatunnel.apache.org/docs/2.3.14/connectors/source/HugeGraph/ +--> + +### 1 版本与兼容性 + +| 组件 | 版本要求 | 说明 | +|------|---------|------| +| Java | 11+ | HugeGraph Client 1.5.0+ 运行环境要求 | +| Apache SeaTunnel | 2.3.13+ | Sink 自 2.3.12 合入；2.3.13 起内置 HugeGraph Client 1.5.0 | +| HugeGraph Server | 1.5.0+ | 需与 Connector 内置 Client 版本匹配 | + +> Source Connector 目前只在 SeaTunnel next（dev）分支，正式版还没带。要用就从源码编译，或等下一个 Release；next 分支内置的 HugeGraph Client 已升到 1.7.0，需搭配对应版本的 Server。 + +### 2 概述 + +[Apache SeaTunnel](https://seatunnel.apache.org/) 是开源数据集成平台，批处理、流式同步都支持，自带 100+ 连接器。HugeGraph 的 Connector-V2 已合入：Sink 负责写入（顶点/边的批量写入、更新、删除），Source 负责读取（批量读取、整图迁移）。 + + + +### 3 选型：SeaTunnel 还是 Loader + +可以把 SeaTunnel 理解为 [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) + [Tools](/cn/docs/quickstart/toolchain/hugegraph-tools) 的合集。两者不冲突：Loader 只管 HugeGraph，开箱即用；SeaTunnel 面向所有数据系统。按场景选： + +| 你的场景 | 推荐 | 理由 | +|---------|------|------| +| 一次性 / 定时把本地文件、HDFS、MySQL 等导入 HugeGraph | Loader | 免部署，映射文件即写即用 | +| 数据已在（或必须经过）Flink、Spark、Kafka 等大数据管道 | SeaTunnel | 复用现有管道，不引入第二套导入工具 | +| 实时流式导入、CDC 增量同步 | SeaTunnel | 原生流式作业 + checkpoint 断点恢复 | +| 导入时自动创建 Schema（PropertyKey / VertexLabel / EdgeLabel） | SeaTunnel | Sink 默认 `CREATE_SCHEMA_WHEN_NOT_EXIST` | +| 只维护 HugeGraph 一张图，规模可控、追求简单 | Loader | 工具链内闭环，无额外集群 | + +选 SeaTunnel 要接受两点：部署一套 SeaTunnel 集群（或复用现有的），作业用通用 HOCON 配置，而不是 HugeGraph 映射文件。 + +### 4 快速开始 + +前置条件：HugeGraph Server 1.5.0+（[部署指南](/cn/docs/quickstart/hugegraph/hugegraph-server)）。 + +#### 4.1 部署 SeaTunnel + +**Docker（推荐）**。拉取镜像，把作业配置所在目录挂载进容器提交，宿主机不用装 Java 环境： + +```bash +docker pull apache/seatunnel:2.3.13 + +docker run --rm -it \ + -v /path/to/job:/config \ + apache/seatunnel:2.3.13 \ + ./bin/seatunnel.sh -e local -c /config/hugegraph-sync.conf +``` + +更多用法见官方 [Docker 部署文档](https://seatunnel.apache.org/docs/getting-started/docker/docker/)。 Review Comment: Fixed by removing the obsolete Docker path. The guide links to the current local deployment documentation and the pinned source-build instructions used for dev. Updated in 1cde03e0fba7cc5a3e76d7a326759eefa874467f. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
