Exec
The exec input plugin, allows to execute external program and collects event logs.
WARNING: Because this plugin invokes commands via a shell, its inputs are subject to shell metacharacter substitution. Careless use of untrusted input in command arguments could lead to malicious command execution.
Container support
This plugin will not function in all the distroless production images as it needs a functional /bin/sh
which is not present. The debug images use the same binaries so even though they have a shell, there is no support for this plugin as it is compiled out.
Configuration Parameters
The plugin supports the following configuration parameters:
Key | Description |
---|---|
Command | The command to execute, passed to popen(...) without any additional escaping or processing. May include pipelines, redirection, command-substitution, etc. |
Parser | Specify the name of a parser to interpret the entry as a structured message. |
Interval_Sec | Polling interval (seconds). |
Interval_NSec | Polling interval (nanosecond). |
Buf_Size | Size of the buffer (check unit sizes for allowed values) |
Oneshot | Only run once at startup. This allows collection of data precedent to fluent-bit's startup (bool, default: false) |
Exit_After_Oneshot | Exit as soon as the one-shot command exits. This allows the exec plugin to be used as a wrapper for another command, sending the target command's output to any fluent-bit sink(s) then exiting. (bool, default: false) |
Propagate_Exit_Code | When exiting due to Exit_After_Oneshot, cause fluent-bit to exit with the exit code of the command exited by this plugin. Follows shell conventions for exit code propagation. (bool, default: false) |
Getting Started
You can run the plugin from the command line or through the configuration file:
Command Line
The following example will read events from the output of ls.
Configuration File
In your main configuration file append the following Input & Output sections:
Use as a command wrapper
To use fluent-bit
with the exec
plugin to wrap another command, use the Exit_After_Oneshot
and Propagate_Exit_Code
options, e.g.:
fluent-bit
will output
then exit with exit code 1.
Translation of command exit code(s) to fluent-bit
exit code follows the usual shell rules for exit code handling. Like with a shell, there is no way to differentiate between the command exiting on a signal and the shell exiting on a signal, and no way to differentiate between normal exits with codes greater than 125 and abnormal or signal exits reported by fluent-bit or the shell. Wrapped commands should use exit codes between 0 and 125 inclusive to allow reliable identification of normal exit. If the command is a pipeline, the exit code will be the exit code of the last command in the pipeline unless overridden by shell options.
Parsing command output
By default the exec
plugin emits one message per command output line, with a single field exec
containing the full message. Use the Parser
directive to specify the name of a parser configuration to use to process the command input.
Security concerns
Take great care with shell quoting and escaping when wrapping commands. A script like
can ruin your day if someone passes it the argument $(rm -rf /my/important/files; echo "deleted your stuff!")'
The above script would be safer if written with:
... but it's generally best to avoid dynamically generating the command or handling untrusted arguments to it at all.
Last updated