plugging in to hadoop

[tejas]

Hi

I am trying to develop a UI and display metrics of mapreduce and HDFS 
and also provide some kind of monitoring.

1. With version 0.20, I was able to extend the 
org.apache.JobTrackerPlugin class by configuring it in the 
mapred-site.properties. Is it possible to make a similar configuration 
in version 0.23? Can i plugin my class to fetch metrics through 
JobTrackerPlugin?
I see that some of the classes in JobTracker class have been depricated? 
So would i have to plugin to the ResourceManager and the NodeManagers to 
get the metrics? Is there any api for that?

2. How does hadoop expose traps and events? If i would have to capture 
events at unfaourable states (like tracker coming down, datanode going 
low on space etc) does Hadoop send out events? How can they be trapped?

Thanks

Tejas