Hi, I am calling JobClient.runJob(conf) function 3-4 times, one after the another in the single node mode in my hadoop program. I have modeled the hadoop changes such that with each call, it gets the same job name so this job basically runs again and again(I am also setting the same JobClient object in each call) With some minor modifications in hadoop JobClient class, it is running the map-reduce phases for each of the call of the JobClient.runJob(conf).
But when I try to run same hadoop program in pseudo distributed mode on this changed hadoop, in first call map reduce operation is performed(it shows map 10% and reduce %) but for the other calls, it shows Running Job: job_22000202 then just after this previous message it directly shows Job Complete:job_22000202. Could some body tell why does it behave so in distributed mode, how can I better debug in distri mode. From what classes hadoop gets the information about previously completed jobs(by job name) so that this status could be changed? Regards, Meera
