The Exec source provides a mechanism to run a command outside Flume and then turn the output into Flume events. To use the Exec source, set the type property to exec:

agent.sources.s1.type=exec

All sources in Flume are required to specify the list of channels to write events to using the channels (plural) property. This is a space-separated list of one or more channel names:

agent.sources.s1.channels=c1

The only other required parameter is the command property, which tells Flume what command to pass to the operating system. Here is an example of the use of this property:

agent.sources=s1
agent.sources.s1.channels=c1
agent.sources.s1.type=exec
agent.sources.s1.command=tail -F /var/log/app.log

Here, I have configured a single source s1 for an agent named agent. The source, an Exec source, will tail the /var/log/app.log file and follow any rotations that outside applications may perform on that log file. All events are written to the c1 channel. This is an example of one of the workarounds...