What is the minimum # of Flume Agents required for very basic implementation?

Hi,

I need your help to understand the following question:

what is the minimum # of Flume Agents required for very basic implementation? For example: Reading a log from Web server & storing it to HDFS.

Regards,
Ravi.

I hope that One Flume-agent is enough to do this… Reading a log from Web server & storing it to HDFS.

That’s a quick response. Appreciate.

where exactly this agent would be running? on the Web server or on the Hadoop Cluster? How does it knows the location of log file on the web server?

Flume agent should be running on the Hadoop Cluster however webserver should communicates with the respective Flume Source.

Do we need to open ports on the web server? if you have working knowledge, can you please explain in detail? I am not able to understand when i read documents on the internet. Appreciate all your help.

You can use netchat or twitter as a source for easy testing/implementation