reporter.progress()

[Curt Holden]

I want to call reporter.progress() from my com.datasalt.pangool.tuplemr.TupleReducer to update the Hadoop framework so long running reducers do not get killed.  However, I cannot figure out how to get an org.apache.hadoop.mapred.Reporter instance from the Pangool API.  Can anyone help with this?

Under the hood we are using Hadoop 0.20.2-cdh3u6.

Thanks,

Curt-

 

 

 

[Curt Holden]

Ah, never mind.  I see that there is a method org.apache.hadoop.mapreduce.TaskInputOutputContext.progress() that I can use.

 

Curt-

[Pere Ferrera]

Hello Curt,

What you need to use is the TupleMRContext.getHadoopContext() over the context you are passed by Pangool into the Reducer or the Mapper -> This will return you the standard Hadoop Context object which you can use normally.

It's probably not very clear in the website, I only found a reference to it here: http://pangool.net/introduction.html

We'll figure out a place in the user guide for clarifying this.

Thanks,