Implementation:Apache Spark Spark Submit Standalone

Field	Value
Source	Doc Spark Standalone
Domains	Deployment
Type	External Tool Doc
Related	Principle:Apache_Spark_Standalone_Job_Submission

Overview

External tool documentation for submitting Spark applications to a standalone cluster via bin/spark-submit.

Description

bin/spark-submit with --master spark://host:port submits applications to the Spark standalone cluster. It supports two deploy modes:

• Client mode (default) -- the driver runs on the machine where spark-submit is invoked
• Cluster mode -- the master launches the driver on one of the worker nodes

In cluster mode, the --supervise flag enables automatic driver restart on failure, providing fault tolerance for long-running applications.

Usage

Use to submit applications to a running standalone cluster. The master URL in spark://host:port format is required. Choose the deploy mode based on the use case:

• Client mode -- for interactive sessions, debugging, and short-lived jobs
• Cluster mode -- for production workloads where the driver should survive client disconnection

Code Reference

Source: docs/submitting-applications.md (L60-175)

Command:

bin/spark-submit --master spark://<host>:<port> [options] <app-jar> [app-args]

Key options:

Option	Description
--master spark://host:port	Required. The master URL for the standalone cluster.
cluster	Where to run the driver (default: client).
--class main-class	The main class of the application.
--supervise	Restart the driver on failure (cluster mode only).

I/O

Direction	Description
Inputs	--master spark://host:port (required), --deploy-mode client or cluster, --class main-class, application JAR, app arguments
Outputs	Application execution results, driver output

Examples

Client mode submission:

./bin/spark-submit \
  --master spark://master:7077 \
  --class com.example.App \
  app.jar

Cluster mode submission:

./bin/spark-submit \
  --master spark://master:7077 \
  --deploy-mode cluster \
  --class com.example.App \
  app.jar

Cluster mode with supervise (auto-restart on failure):

./bin/spark-submit \
  --master spark://master:7077 \
  --deploy-mode cluster \
  --supervise \
  --class com.example.App \
  app.jar

Related

• Principle:Apache_Spark_Standalone_Job_Submission