I has been using flume to ingest data into HBase for almost a year. Everything works out fine.
I wrote my own implemetation of AsyncHbaseEventSerializer to generate Puts.
However, I found there is a real limitation here. I can't serialize something into DeleteRequest. AsyncHbaseEventSerializer does not have such method.
So how do you suppose to use flume for cleanup purpose?
Day 1: I put two columns for a row c1=a; c2=b
Day 2: I got a new record for the row c1=a_chg
My Serializer can easily generate put request to put c1=a_chg; But c2 is removed and I need to delete column c2 for this rowKey. I can't figure out how to generate a DeleteRequest.
What is your experience? What should I do?