SystemExecute
Short Description
Functionality of the SystemExecute component has been superseded by the ExecuteScript component. If you wish to run scripts as part of you job, use a jobflow with the ExecuteScript component. |
SystemExecute executes system commands.
Same input metadata | Sorted inputs | Inputs | Outputs | Each to all outputs[1] | Java | CTL | Auto-propagated metadata |
---|---|---|---|---|---|---|---|
- |
⨯ |
0-1 |
0-1 |
- |
⨯ |
⨯ |
⨯ |
Ports
Port type | Number | Required | Description | Metadata |
---|---|---|---|---|
Input |
0 |
⨯ |
For standard input of the specified system command (input of the process) |
Any1 |
Output |
0 |
For a standard output of the specified system command (output of the process) |
Any2 |
Metadata
SystemExecute does not propagate metadata.
SystemExecute has no metadata template.
SystemExecute Attributes
Attribute | Req | Description | Possible values |
---|---|---|---|
Basic |
|||
System command |
yes |
A command to be executed by the system. The command is always saved to a tmp file as a script. In case an interpreter is specified, it executes that script. If the command requires an input, it has to be sent to the command through the optional input port. For details, see How it works. |
|
Process input/output charset |
Encoding used for formatting/parsing data for input/from output of system process. The default encoding depends on DEFAULT_CHARSET_DECODER in defaultProperties. |
UTF-8 | <any other> |
|
Output file URL |
Name of a file, including the path, to which output of the process (together with errors) should be written if the output edge is not connected and if System command creates the standard output. For more information, see URL File Dialog. |
||
Append |
By default, the contents of an output file is always deleted and overwritten by new data.
If set to |
false (default) | true |
|
Command interpreter |
An interpreter that should execute the command.
If specified, System command is saved as a script to a temporary file and executed by this interpreter.
Its form must be the following: See the list of well known interprets. |
||
Working directory |
A working directory of the component. |
current directory[2] (default) | other directory |
|
Advanced |
|||
Number of error lines |
The number of lines that are printed if a command finishes with errors. |
2 (default) | 1-N |
|
Environment |
System-dependent mapping from variables to values.
Mappings are separated by a colon, semicolon or pipe.
By default, the new value is appended to the environment of the current process. Both |
For example: PATH=/home/user/mydir[!true] (default) | PATH=/home/user/mydir!false |
|
Timeout for producer/consumer workers (ms) |
Timeout; by default in milliseconds, but other time units may be used. For details, see Timeout. |
0 (without limitation) | 1-n |
|
Ignore exit value |
In case the executed system command returns non-zero value, the component fails. This option can change this behavior, the exit value can be ignored. |
true | false (default) |
Details
SystemExecute executes commands and arguments specified in the component itself as a separate process. The commands receive a standard input through the input port and send a standard output to the output port (if the command creates any output).
How it works
SystemExecute runs the command specified in the System command and creates two threads.
-
The first thread (producer) reads records from the input edge, serializes them and sends them to
stdin
of the command. -
The second thread (consumer) reads
stdout
of the command, parses it and sends it to the output edge.
Timeout
-
When the command ends, the component still waits until both the producer and the consumer also finish their work. The time is defined in the Timeout attribute.
-
By default, timeout is unlimited. In the case of an unexpected deadlock, you can set the timeout to any number of milliseconds.
Well known command interprets
The following list contains possible interprets that can be used in SystemExecute component. You are not limited to the items from this list.
Windows
-
cmd /c ${}
-
powershell ${}
Linux
-
/bin/sh ${}
-
/bin/bash ${}
-
/bin/tcsh ${}
-
/usr/bin/perl ${}
-
/usr/bin/python ${}
Difference between SystemExecute and ExecuteScript
SystemExecute uses a data-oriented approach. It allows you to stream data from and to a script.
ExecuteScript executes scripts in steps. It allows you to receive scripts from an input edge and execute them one by one.
Examples
Run command on Linux
This example shows execution of command on Linux using SystemExecute.
Run the uptime
command using the SystemExecute component.
Send the results through an edge to the next component.
Solution
Connect the output port of SystemExecute with the next component.
Metadata of the edge should contain just one string
field.
Configure SystemExecute.
Attribute | Value |
---|---|
System command |
uptime |
Command interpreter |
/bin/bash ${} |
Run command on Windows
This example shows execution of command on MS Windows using SystemExecute
Run tasklist
using SystemExecute component.
Send the results trough an edge to the next component.
Solution
Connect the output port of SystemExecute with the next component.
Metadata of the edge should contain just one string
field.
Configure SystemExecute.
Attribute | Value |
---|---|
System command |
tasklist |
Process input/output charset |
windows-1252 |
Command interpreter |
cmd /c ${} |
If process input/output charset is not set or it is set incorrectly, you can encounter the following error message in the graph log: Character decoding error occurred. Set correct charset. Current charset is UTF-8. |
Run command that saves its output to file
This example shows the way to run a command with SystemExecute and save the output from the command to a file.
Run command who
and save its output to the file logged_users.txt
for further processing.
Solution
Attribute | Value |
---|---|
System command |
who |
Output file URL |
${DATAOUT_DIR}/logged_users.txt |
Command interpreter |
/bin/bash ${} |
Calling external filter
This example shows the way to use an external shell filter from SystemExecute: the data from a graph is sent to input stream of the filter and the output of the script is sent back to the graph.
Call sed
to replace A
with B
.
Solution
Connect the input port of SystemExecute with the component producing the data and the output port of SystemExecute with the component consuming the data. Assign metadata to the edges. The metadata on both edges should be the same, unless the script changes the number of columns or delimiters.
Attribute | Value |
---|---|
System command |
sed 's_A_B_g' |
Command interpreter |
/bin/bash ${} |
Best Practices
We recommend users to explicitly specify Process input/output charset.
Compatibility
Version | Compatibility Notice |
---|---|
6.4 |
SystemExecute was deprecated. Use ExecuteScript instead. |