Thanks for the suggestions. But What I want to do is aggregate all the logs from the cluster to one specific node, so that I always know where they are.

I do not want to go to every node in my cluster, and check the content of the log. I am not using Hadoop, so I will not have WEB access to hdfs, so that I can access logs from there..

My system has Kafka, Spark & Cassandra. Spark will work on Mesos.