klout/brickhouse

CollectUDAF makes the tasks fail because of out of memory

XiaolZhang opened this issue · 0 comments

I use 'brickhouse.udf.collect.CollectUDAF' to create function collect_map, the origin data is 3 billions and the aggregrated data is 593 millions,however, the reduce tasks will always fail with throwing ioexecption, what's the problem?
image