3

I am trying to find a tool, which would show me an overview of my Hadoop ecosystem - state, health, running tasks, etc. I tried to Google, but did not find any. Is there some nice useful tool?

Vojtech
  • 31
  • 2

3 Answers3

1

Running tasks, mapred node state/health:

http://example.com:50030/jobtracker.jsp

HDFS health and node state:

http://example.com:50070/dfshealth.jsp

Replace example.com with the IP or hostname of your job tracker / name node.

If those ports aren't correct check your mapred-site.xml and hdfs-site.xml for the defined port.

Dan R
  • 2,335
  • 2
  • 19
  • 28
0

Cloudera offers a proprietary tool which does what you describe and more called Cloudera Manager. It isn't F/OSS, but they do have a free edition that lets you run on some limited number of nodes. It's meant as a full lifecycle product, so it might be a bit overkill for what you need.

Full disclosure: I've never used it myself. We considered it at my company, but ended up not using it.

Scrivener
  • 3,116
  • 1
  • 22
  • 24
0

Another alternative is the Ambari project. It's currently in incubation status at Apache, but I believe Hortonworks is one of the major drivers for it. It's primarily a provisioning and monitoring framework. You can see examples at Hortonworks' website. They did a blog post describing how to set it up.

If you're looking for a "prettier" interface to the jobtracker or namenode, Cloudera has the Hadoop User Environment (HUE), which provides some of the functionality you're looking for. It's geared towards non-administrators for the most part, though.

Travis Campbell
  • 1,466
  • 7
  • 15