Hadoop dashboard27 Oct 2016
Now we have dashboard for out hadoop clusters. On github you can see sourcecode (link is bottom).
At sourcecode are (for Hadoop) the most important jobs
- hdfs_status.rb - get count live, dead and decommission nodes
- yarn_status.rb - get count running, lost and unhealth nodemanagers and running/failing jobs (more info bottom)
- hbase_status.rb - get count live regionservers, dead regionservers and region in transitions
Here you can see cluster with HDFS + YARN + HBase.
In next screenshot you can see failing a running jobs. This jobs are download and parse from
Failing jobs shows:
- datetime when job failed
- last success complete job with same name
Running jobs shows:
- datetime start
- type (yarn, spark, mr, ...)