The data volumes in the millions, how should solve? Increase Mr Is

Error: GC overhead limit exceeded
17/05/27 15:47:25 INFO graphs. Job: Task Id: attempt_149555519_0057_m_000000_1, Status: FAILED
Error: GC overhead limit exceeded
17/05/27 15:48:16 INFO graphs. Job: Task Id: attempt_149555519_0057_m_000000_2, Status: FAILED
Error: GC overhead limit exceeded
17/05/27 15:49:10 INFO graphs. The Job: the map reduce 100% 0%
17/05/27 15:49:10 INFO graphs. The Job: the Job job_149555519_0057 failed with the state failed due to: Task failed task_149555519_0057_m_000000
The Job failed as the tasks failed. FailedMaps: 1 failedReduces: 0

17/05/27 15:49:10 INFO graphs. Job: Counters: 11
Job Counters
Failed the map tasks=4
Launched the map tasks=4
Other local map tasks=4
The Total time spent by all maps in occupied slots (ms)=197894
The Total time spent by all reduces occupied in slots (ms)=0
The Total time spent by all the map tasks (ms)=197894
Total vcore - seconds seems by all the map tasks=197894
Total Megabyte - seconds seems by all the map tasks=202643456
The Map - Reduce Framework
The CPU time spent (ms)=0
Physical memory (in bytes) snapshot=0
Virtual memory (in bytes) snapshot=0
17/05/27 15:49:10 WARN graphs. Counters: Group FileSystemCounters is deprecated. Use org. Home. The apache hadoop. Graphs. FileSystemCounter
17/05/27 15:49:10 INFO graphs. ImportJobBase: Transferred 0 bytes in 214.2624 seconds (0 bytes/SEC)
17/05/27 15:49:10 INFO graphs. ImportJobBase: Retrieved 0 records.
17/05/27 15:49:10 ERROR tool. ImportTool: ERROR during import, an import job failed