In the past, not knowing any better, I've made dummy imputs. Make a file of as many lines as you'd like mappers, use TextInputFormat and just ignore the input or trigger what the particular mapper does off passed input.
St.Ack On Tue, Jul 21, 2009 at 8:13 PM, bharath vissapragada < [email protected]> wrote: > By saying "using API in your job to pull from multiple table" do you mean > that in the Map phase itself we use HBase API to fetch table rows ? If yes > then what should we give as input to the map function , can we leave it as > a > blank? > > On Wed, Jul 22, 2009 at 12:43 AM, Jonathan Gray <[email protected]> wrote: > > > Currently, there is not. > > > > You would have multiple MR jobs, or you would directly use the API in > your > > job to pull from multiple tables. > > > > I suppose it would be feasible, but as it is now, you are not told which > > table your Result comes from. > > > > > > bharath vissapragada wrote: > > > >> Hi all, > >> > >> > >> Generally TableMap.initJob() method takes a "table name" as input to > >> the > >> map while using map-reduce in HBase . > >> > >> Is there a way so that i can use more than 1 table , i.e., input to the > >> map > >> contains more than 1 table , > >> > >> Thanks > >> > >> >
