02-19-2015 03:27 AM
As part of my setup, I have a number of scripts for input ingestion and oozie tasks, which as part of their processing do their fair share of info/warn/error logging in languages such as python,java, R.
I would like to have some centralized way to check out the errors on my pipeline, but would like a lightweight solution (i.e. I know splunk but afraid it may be an overkill for this). Is there some simpler solution, where I could leverage cloudera manager to analyze my applicational logs ? (notice that some of the ingestion scripts running on edge nodes, may even be simple "cron'ed" scripts and not integrated in the hadoop pipeline..
Even if I used flume to ingest logs, which would be the "lightest" application where I could analyze them ?