goutamadwant · Jul 3, 2026 · Jul 4, 2026 · Jul 4, 2026 · Jul 5, 2026 · Jul 7, 2026
diff --git a/.github/workflows/backend.yml b/.github/workflows/backend.yml
@@ -712,6 +712,10 @@ jobs:
           # flags must be repeated here as well as in the pom.xml surefire.module.args property.
           SUREFIRE_JVM_ARGS: -Xmx3072m -XX:MaxMetaspaceSize=512m -Dfile.encoding=UTF-8 --add-opens=java.base/java.lang=ALL-UNNAMED --add-opens=java.base/java.net=ALL-UNNAMED --add-opens=java.base/java.nio=ALL-UNNAMED --add-opens=java.base/java.util=ALL-UNNAMED --add-opens=java.base/sun.nio.ch=ALL-UNNAMED --add-exports=java.security.jgss/sun.security.krb5=ALL-UNNAMED
         if: runner.os == 'Windows'
+      - name: Run packaged Spark 3.5 launcher tests
+        run: ./mvnw -B -T 1 verify -pl seatunnel-core/seatunnel-spark-starter/seatunnel-spark-3.5-starter -am -DskipUT=true -DskipIT=false -D"it.test=TestSpark35StarterJarContentIT" -D"it.failIfNoSpecifiedTests=false" -D"license.skipAddThirdParty"=true -D"skip.ui"=true --no-snapshot-updates
+        env:
+          MAVEN_OPTS: -Xmx1536m -Dfile.encoding=UTF-8
       - name: Upload test reports
         if: failure()
         uses: ./.github/actions/upload-test-reports

diff --git a/docs/en/engines/command/usage.mdx b/docs/en/engines/command/usage.mdx
@@ -27,6 +27,8 @@ bin/start-seatunnel-spark-2-connector-v2.sh
 bin/start-seatunnel-spark-3-connector-v2.sh
 ```
 
+For Spark 3.5.x, use `bin/start-seatunnel-spark-3.5-connector-v2.sh` with the same options. Set `SPARK_HOME` to the matching Spark installation. See [Spark engine compatibility](../spark.md) for the current validation limits.
+
 </TabItem>
 <TabItem value="flink13">
 

diff --git a/docs/en/engines/spark.md b/docs/en/engines/spark.md
@@ -10,6 +10,46 @@ Use this path if you want to run SeaTunnel on Spark:
 - [Quick Start With Spark](../getting-started/locally/quick-start-spark.md)
 - [Job Configuration Guide](../getting-started/job-configuration-guide.md)
 
+SeaTunnel provides separate Spark 3 starters so existing Spark 3.3 deployments can remain on
+their current runtime while Spark 3.5 deployments use the Spark 3.5.8 starter:
+
+- `seatunnel-spark-3-starter.jar` for Spark 3.3.x
+- `seatunnel-spark-3.5-starter.jar` for Spark 3.5.x
+
+For an existing Spark 3.4 deployment, retain the existing `seatunnel-spark-3-starter.jar`
+and `start-seatunnel-spark-3-connector-v2` launcher. This change does not add a separate
+Spark 3.4 starter or extend its compatibility coverage.
+
+The Spark 3.5 starter reuses the translation layer compiled against Spark 3.3.
+Row-encoder, batch-write and bounded micro-batch sink tests exercise selected paths on Spark 3.5.8, but do not
+prove compatibility of every Catalyst or DataSource V2 path. Other binary
+incompatibilities may surface only when those paths execute on the target runtime.
+Validate your connectors and deployment mode before upgrading.
+
+The public Spark 3.5 starter uses the batch `read()` and `write().save()` pipeline;
+use a batch job configuration with this launcher. The micro-batch test runs a finite
+Spark file stream through the row encoder and SeaTunnel streaming sink adapter. It
+does not validate the SeaTunnel streaming source, checkpoint restore, or streaming
+jobs submitted through the public starter. Spark 3.5 connector-container tests are
+scheduled-only (`testInPR=false`); normal PR CI does not run that runtime matrix.
+
+The Spark 3.5 Windows `.cmd` launcher still passes a command string to `spark-submit.cmd`;
+it does not use the Unix launcher's argument-file protocol. Arbitrary arguments
+containing embedded quotes or Windows command-shell metacharacters are not
+supported, and native Windows launcher behavior has not been validated. Do not
+pass untrusted values through this launcher; use the Unix launcher when exact
+argument preservation is required.
+
+The packaged Windows tests exercise help, JVM startup failure, and opening quoted
+configuration paths containing spaces, `!`, and `&`. The path tests stop at a
+configuration parse error, before Spark submission; they do not establish exact
+argument preservation by `spark-submit.cmd`. Native Windows CI must pass on the
+revision being validated before treating these launcher cases as verified.
+
+The Unix packaged test runs the real starter JVM and checks argument preservation
+for paths containing spaces and both quote characters. It captures the arguments
+at `spark-submit`; it does not submit a Spark job.
+
 ## When To Choose Spark
 
 Spark is usually the right engine when:
@@ -49,6 +89,8 @@ Spark on YARN client mode:
 ./bin/start-seatunnel-spark-3-connector-v2.sh --master yarn --deploy-mode client --config config/example.conf
 ```
 
+For Spark 3.5, use `start-seatunnel-spark-3.5-connector-v2.sh` with the same options.
+
 ## Minimal Example Job
 
 The example below runs on Spark and prints generated records to the console.

diff --git a/docs/en/getting-started/docker/docker.md b/docs/en/getting-started/docker/docker.md
@@ -132,6 +132,8 @@ docker run --rm -it apache/seatunnel bash ./bin/start-seatunnel-spark-2-connecto
 docker run --rm -it apache/seatunnel bash ./bin/start-seatunnel-spark-3-connector-v2.sh -c config/v2.batch.config.template
 ```
 
+For Spark 3.5.x, use `start-seatunnel-spark-3.5-connector-v2.sh` instead. The container must include Spark 3.5.x, the matching SeaTunnel starter, and `SPARK_HOME` pointing to that Spark installation; changing the launcher name alone does not upgrade the image. See [Spark engine compatibility](../../engines/spark.md) for validation limits.
+
 - Flink
   before you submit job, you need start flink cluster first.
 

diff --git a/docs/en/getting-started/locally/quick-start-spark.md b/docs/en/getting-started/locally/quick-start-spark.md
@@ -27,7 +27,12 @@ see [Getting Started: Standalone](https://spark.apache.org/docs/latest/spark-sta
 
 ## Step 3: Add Job Config File To Define A Job
 
-Edit `config/v2.streaming.conf.template`, which determines the way and logic of data input, processing, and output after seatunnel is started.
+For Spark 3.5, edit `config/v2.batch.config.template` with `job.mode = "BATCH"`.
+The bounded Spark 3.5.8 micro-batch test covers the row encoder and streaming sink
+adapter. Streaming through the public starter and checkpoint restore remain
+unvalidated, so the streaming template is not a verified Spark 3.5 example. Existing
+Spark 2.4/3.3 deployments can continue
+using `config/v2.streaming.conf.template` as shown below.
 The following is an example of the configuration file, which is the same as the example application mentioned above.
 
 ```hocon
@@ -89,7 +94,7 @@ cd "apache-seatunnel-${version}"
 --config ./config/v2.streaming.conf.template
 ```
 
-Spark3.x.x
+Spark 3.3.x
 
 ```shell
 cd "apache-seatunnel-${version}"
@@ -99,6 +104,26 @@ cd "apache-seatunnel-${version}"
 --config ./config/v2.streaming.conf.template
 ```
 
+Spark 3.4.x
+
+For an existing Spark 3.4 deployment, retain `seatunnel-spark-3-starter.jar` and the
+`start-seatunnel-spark-3-connector-v2` launcher shown above. This change adds no
+separate Spark 3.4 starter and no new Spark 3.4 compatibility validation.
+
+Spark 3.5.x
+
+Start with a batch job. The bounded micro-batch sink-adapter test does not validate
+streaming through the public starter or checkpoint restore; the streaming template
+is not part of this quick start for Spark 3.5.
+
+```shell
+cd "apache-seatunnel-${version}"
+./bin/start-seatunnel-spark-3.5-connector-v2.sh \
+--master local[4] \
+--deploy-mode client \
+--config ./config/v2.batch.config.template
+```
+
 **See The Output**: When you run the command, you can see its output in your console. This
 is a sign to determine whether the command ran successfully or not.
 

diff --git a/docs/zh/engines/command/usage.mdx b/docs/zh/engines/command/usage.mdx
@@ -27,6 +27,8 @@ bin/start-seatunnel-spark-2-connector-v2.sh
 bin/start-seatunnel-spark-3-connector-v2.sh
 ```
 
+对于 Spark 3.5.x，请使用 `bin/start-seatunnel-spark-3.5-connector-v2.sh`，选项相同。将 `SPARK_HOME` 指向匹配的 Spark 安装目录。当前验证范围请参阅 [Spark 引擎兼容性](../spark.md)。
+
 </TabItem>
 <TabItem value="flink13">
 

diff --git a/docs/zh/engines/spark.md b/docs/zh/engines/spark.md
@@ -10,6 +10,39 @@
 - [Spark 引擎快速开始](../getting-started/locally/quick-start-spark.md)
 - [作业配置指南](../getting-started/job-configuration-guide.md)
 
+SeaTunnel 提供独立的 Spark 3 启动器。现有 Spark 3.3 部署可以继续使用原有运行时，
+Spark 3.5 部署则可以使用 Spark 3.5.8 启动器：
+
+- `seatunnel-spark-3-starter.jar` 用于 Spark 3.3.x
+- `seatunnel-spark-3.5-starter.jar` 用于 Spark 3.5.x
+
+现有 Spark 3.4 部署应继续使用原来的 `seatunnel-spark-3-starter.jar` 和
+`start-seatunnel-spark-3-connector-v2` 启动脚本。本次变更不新增独立的 Spark 3.4
+启动器，也不扩展其兼容性验证范围。
+
+Spark 3.5 启动器复用了基于 Spark 3.3 编译的转换层。行编码器、批量写入和有界微批写入测试在
+Spark 3.5.8 上验证了部分路径，但不能证明所有 Catalyst 或 DataSource V2 路径均兼容。
+其他二进制兼容性问题可能仅在目标运行时执行相关路径时暴露。升级前请验证所用连接器
+和部署模式。
+
+Spark 3.5 公共启动器使用批处理的 `read()` 和 `write().save()` 执行路径，应使用批处理作业配置。
+微批测试将有限的 Spark 文件流通过行编码器写入 SeaTunnel 流式 Sink 适配器，不覆盖 SeaTunnel
+流式 Source、检查点恢复，或通过公共启动器提交流处理作业。Spark 3.5 连接器容器测试仅在定时
+任务中运行（`testInPR=false`），普通 PR CI 不运行该运行时矩阵。
+
+Spark 3.5 Windows `.cmd` 启动器仍将命令字符串传递给 `spark-submit.cmd`，未采用 Unix
+启动器的参数文件协议。不支持包含嵌入引号或 Windows 命令解释器元字符的任意参数，
+且尚未验证原生 Windows 启动器行为。请勿通过此启动器传入不可信的值；需要精确保留
+参数时，请使用 Unix 启动器。
+
+Windows 打包测试覆盖帮助输出、JVM 启动失败，以及打开包含空格、`!` 和 `&` 的带引号
+配置路径。路径测试在配置解析错误处停止，不会提交 Spark 作业，因此不能证明
+`spark-submit.cmd` 会精确保留参数。只有待验证版本的原生 Windows CI 通过后，
+才能将这些启动器场景视为已验证。
+
+Unix 打包测试运行真实的启动器 JVM，检查包含空格及单、双引号的路径参数是否完整保留。
+该测试在 `spark-submit` 边界捕获参数，不会提交 Spark 作业。
+
 ## 什么时候选择 Spark
 
 以下场景通常更适合使用 Spark：
@@ -49,6 +82,8 @@ Spark on YARN 客户端模式：
 ./bin/start-seatunnel-spark-3-connector-v2.sh --master yarn --deploy-mode client --config config/example.conf
 ```
 
+使用 Spark 3.5 时，请使用 `start-seatunnel-spark-3.5-connector-v2.sh`，其他参数保持不变。
+
 ## 最小示例作业
 
 下面这个例子会在 Spark 上运行，并把生成的数据打印到控制台。

diff --git a/docs/zh/getting-started/docker/docker.md b/docs/zh/getting-started/docker/docker.md
@@ -132,6 +132,8 @@ docker run --rm -it apache/seatunnel bash ./bin/start-seatunnel-spark-2-connecto
 docker run --rm -it apache/seatunnel bash ./bin/start-seatunnel-spark-3-connector-v2.sh -c config/v2.batch.config.template
 ```
 
+对于 Spark 3.5.x，请改用 `start-seatunnel-spark-3.5-connector-v2.sh`。容器必须包含 Spark 3.5.x 和匹配的 SeaTunnel 启动器，并将 `SPARK_HOME` 指向该 Spark 安装目录；仅更改启动器名称不会升级镜像。验证范围请参阅 [Spark 引擎兼容性](../../engines/spark.md)。
+
 - Flink
   在提交作业之前，您需要先启动 Flink 集群。
 

diff --git a/docs/zh/getting-started/locally/quick-start-spark.md b/docs/zh/getting-started/locally/quick-start-spark.md
@@ -27,7 +27,10 @@ sidebar_position: 4
 
 ## 步骤 3: 添加作业配置文件来定义作业
 
-编辑`config/v2.streaming.conf.template`，它决定了当SeaTunnel启动后数据输入、处理和输出的方式及逻辑。
+Spark 3.5 请编辑 `config/v2.batch.config.template`，并设置 `job.mode = "BATCH"`。
+Spark 3.5.8 的有界微批测试覆盖行编码器和流式 Sink 适配器。通过公共启动器执行流处理和
+检查点恢复仍未验证，因此流处理模板不是已验证的 Spark 3.5 示例。
+现有 Spark 2.4/3.3 部署可以继续使用下文所示的 `config/v2.streaming.conf.template`。
 下面是配置文件的示例，它与上面提到的示例应用程序相同。
 
 ```hocon
@@ -89,7 +92,7 @@ cd "apache-seatunnel-${version}"
 --config ./config/v2.streaming.conf.template
 ```
 
-Spark 3.x.x
+Spark 3.3.x
 
 ```shell
 cd "apache-seatunnel-${version}"
@@ -99,6 +102,25 @@ cd "apache-seatunnel-${version}"
 --config ./config/v2.streaming.conf.template
 ```
 
+Spark 3.4.x
+
+现有 Spark 3.4 部署应继续使用 `seatunnel-spark-3-starter.jar` 和上文所示的
+`start-seatunnel-spark-3-connector-v2` 启动脚本。本次变更不新增独立的 Spark 3.4
+启动器，也不新增 Spark 3.4 兼容性验证。
+
+Spark 3.5.x
+
+请先运行批处理作业。有界微批 Sink 适配器测试不验证通过公共启动器执行流处理或
+检查点恢复，因此 Spark 3.5 的快速开始不使用流处理模板。
+
+```shell
+cd "apache-seatunnel-${version}"
+./bin/start-seatunnel-spark-3.5-connector-v2.sh \
+--master local[4] \
+--deploy-mode client \
+--config ./config/v2.batch.config.template
+```
+
 **查看输出**: 当您运行该命令时，您可以在控制台中看到它的输出。您可以认为这是命令运行成功或失败的标志。
 
 SeaTunnel控制台将会打印一些如下日志信息:

diff --git a/pom.xml b/pom.xml
@@ -102,6 +102,7 @@
         <flink.1.20.1.version>1.20.1</flink.1.20.1.version>
         <spark.2.4.0.version>2.4.0</spark.2.4.0.version>
         <spark.3.3.0.version>3.3.0</spark.3.3.0.version>
+        <spark.3.5.8.version>3.5.8</spark.3.5.8.version>
         <spark.binary.2.4.version>2.4</spark.binary.2.4.version>
         <commons.beanutils.version>1.9.4</commons.beanutils.version>
         <commons.cli.version>1.4</commons.cli.version>

diff --git a/seatunnel-core/seatunnel-spark-starter/pom.xml b/seatunnel-core/seatunnel-spark-starter/pom.xml
@@ -33,6 +33,7 @@
     <modules>
         <module>seatunnel-spark-2-starter</module>
         <module>seatunnel-spark-3-starter</module>
+        <module>seatunnel-spark-3.5-starter</module>
         <module>seatunnel-spark-starter-common</module>
     </modules>
 
@@ -71,11 +72,12 @@
                                     jul-to-slf4j
                                     jcl-over-slf4j
 
-                                Spark(3.x) server lib already include:
+                                Spark(3.x) server lib already include (binding varies by version):
                                     slf4j-api
                                     log4j-api
                                     log4j-core
                                     log4j-slf4j-impl
+                                    log4j-slf4j2-impl
                                     log4j-1.2-api
                                     jul-to-slf4j
                                     jcl-over-slf4j
@@ -96,7 +98,15 @@
                             <exclude>ch.qos.logback:*</exclude>
                             <exclude>org.apache.logging.log4j:log4j-api</exclude>
                             <exclude>org.apache.logging.log4j:log4j-core</exclude>
+                            <!--
+                                These are shade exclusions, not starter/logging includes.
+                                The launcher JVM uses SeaTunnel's SLF4J 1.7 API and
+                                log4j-slf4j-impl from starter/logging. Spark submission
+                                starts a separate JVM using Spark's own logging libraries;
+                                neither binding should be embedded in the starter jar.
+                            -->
                             <exclude>org.apache.logging.log4j:log4j-slf4j-impl</exclude>
+                            <exclude>org.apache.logging.log4j:log4j-slf4j2-impl</exclude>
                             <!-- spark3.x use slf4j + log4j2.x -->
                             <exclude>org.apache.logging.log4j:log4j-to-slf4j</exclude>
                             <exclude>org.apache.seatunnel:seatunnel-shade-hadoop3-uber</exclude>
-Original file line number
+Diff line change
@@ Expand Up @@
     docker run --rm -it apache/seatunnel bash ./bin/start-seatunnel-spark-3-connector-v2.sh -c config/v2.batch.config.template
     ```
+    For Spark 3.5.x, use `start-seatunnel-spark-3.5-connector-v2.sh` instead. The container must include Spark 3.5.x, the matching SeaTunnel starter, and `SPARK_HOME` pointing to that Spark installation; changing the launcher name alone does not upgrade the image. See [Spark engine compatibility](../../engines/spark.md) for validation limits.
     - Flink
       before you submit job, you need start flink cluster first.
@@ Expand Down @@