Script jobs-SchedulerX(SchedulerX)-阿里云帮助中心

How it works

The following diagram shows how script jobs work.

yuque_diagram (1)

To use script jobs, you must deploy the schedulerx-agent on your target machines. The agent communicates with the server, receives script execution commands, and returns the execution results.
Install all required script dependencies, such as Python libraries or other files, on the target machines.
When a script executes, schedulerx-agent forks a child process to run it. This child process consumes memory on the host machine, not from the schedulerx-agent itself. The amount of memory consumed depends on the script's operations. Running many script jobs simultaneously can exhaust the machine's memory. For resource-intensive scripts, we recommend using a Kubernetes job to run them in a pod.

Prerequisites

To run scripts, the schedulerx-agent must be deployed.

Create a script job

Log on to the SchedulerX console. In the left-side navigation pane, navigate to the Task Management page and create a job. Set the Job Type to script job.

Enter a Task Name and Description, select an Application ID, set the priority (Medium by default), and enter the required parameters in the Task Parameters field.

Step 1: Select a script language

Select a Script Language. SchedulerX supports Shell, Python, and PHP.

Step 2: Execution command

Specify the command to run the script for different languages:

Shell script: Defaults to /bin/sh.
Python script: Defaults to python. You can change this to python2 or python3.
PHP script: Defaults to php.

Step 3: File format

The unix and windows file formats use different line endings.

If the script will run on a Windows system, select windows.
If the script will run on a Linux/Unix/Mac system, select unix.

Important

If you author the script on a Windows system but run it on a Linux system, set File format to unix.

Step 4: Execution mode

Prerequisite: The schedulerx-agent must be deployed. For more information, see Connect an agent (for Script or HTTP jobs).

Execution Mode:

Stand-alone operation: SchedulerX randomly selects one of your deployed agents to run the script.
Broadcast run: SchedulerX runs the script on all deployed agents simultaneously. This mode is used for batch operations and maintenance.
Shard run: Distributes the job execution across multiple agents. For more information, see Create a Python sharding job.

Write and edit the script

After you create a script job, SchedulerX generates a default script template based on the language you selected. On the Edit Script page, click Edit Script to modify the script.

The job list shows each job's Execution mode (such as Stand-alone operation), and provides an EditScript button for modifications.

Manage script versions

The Edit Script includes a script version history feature. On the EditScript page, enable the Historical versions switch to open the version comparison view.

The left pane shows the current version, and the right pane shows a historical version.
Select a version from the drop-down list to view its details, including the version notes and creation time.
After modifying the script, click Confirm and enter a new version name when prompted.

Demo

Broadcast run

Connect two schedulerx-agent instances.

In the Connected Instances panel, verify that both instances (e.g., 172.16.13.125 and 172.16.13.120) are connected as version 1.12.5 pods and have a health status of Healthy.

Create a script job with Shell as the Script Language and Execution Mode as the Execution mode. Use the following script content:

#!/bin/bash
# #{schedulerx.shardingId} is the shard ID
echo "shard ID: #{schedulerx.shardingId}"
# #{schedulerx.shardingNum} is the total number of shards
echo "total shards: #{schedulerx.shardingNum}"
echo "hello world"
exit 0

In the Actions column for the target job, click Run once.

Navigate to the Execution List page. Find the job instance and click Log in the Operation column to view the output from each machine.

The log for 172.16.13.125 shows that the shard ID is 1 and the total shards is 2.

time : 2025-04-22 17:05:26
---
ip : 172.16.13.125:34487
executionId : 10496431_74039165530_1
level : INFO
log : total shards: 2
time : 2025-04-22 17:05:26
---
ip : 172.16.13.125:34487
executionId : 10496431_74039165530_1
level : INFO
log : hello world
time : 2025-04-22 17:05:26
---
ip : 172.16.13.125:34487
executionId : 10496431_74039165530_1
level : INFO
log : shard ID: 1
time : 2025-04-22 17:05:26

The log for 172.16.13.120 shows that the shard ID is 0 and the total shards is 2.

Shard run

Create a script job with Shell as the Script Language and Execution Mode as the Execution mode. Configure the other job settings as follows:

Set Execution command to /bin/sh, File format to unix, Executor to Agent, Priority to Medium, and Sharding parameter to 0=hangzhou,1=shanghai,2=beijing,3=shenzhen.

Once the job is created, click its Edit Script button and enter the following script content:

#!/bin/bash
# #{schedulerx.shardingId} is the shard ID
echo "shard ID: #{schedulerx.shardingId}"
# #{schedulerx.shardingParameters} is the sharding parameter
echo "sharding parameter: #{schedulerx.shardingParameters}"
echo "hello world"
exit 0

In the Actions column for the target job, click Run once.
Navigate to the Execution List page. Find the job instance and click Log in the Operation column to view the sharding parameter assigned to each machine.
- The logs show that 172.16.13.120 ran twice, processing the hangzhou and beijing shards.
- The logs show that 172.16.13.125 also ran twice, processing the shenzhen and shanghai shards.