Hi, Thanks ted. we are using default split policy and our flush size is 64 MB. And the size is calculated based on the formula
Math.min(getDesiredMaxFileSize(),initialSize * tableRegionsCount * tableRegionsCount * tableRegionsCount); If this size exceeds max region size (10 GB), then max region size will be taken or else calculated value will be taken. Suppose if the calculation returns 11 gb then 10 gb will be considered and it will be sent to mapper. Default mapper mapred.map.child.java.opts is 1 gb. what my doubt what if 10gb of data is sent to this mapper with 1 gb whether there will be any issue?? Thanks On Mon, May 15, 2017 at 2:22 PM, Ted Yu <[email protected]> wrote: > Split policy may play a role here. > > Please take a look at: > http://hbase.apache.org/book.html#_custom_split_policies > > On Mon, May 15, 2017 at 1:48 AM, Rajeshkumar J < > [email protected]> > wrote: > > > Hi, > > > > As we run mapreduce over hbase it will take each region as input for > each > > mapper. I have given region max size as 10GB. If i have about 5 gb will > it > > take 5 gb of data as input of mappers?? > > > > Thanks > > >
