we deploy hdfs in our company. we meet a unormal situation.
we set heap memory to 280G, but we actually consume 450G.
*we seen this by pmap.*
* 2ae91c000000 rw-p 00000000 00:00 0 294174720 293705824
293705824 293705824 293705824 0 0 0198c000 rw-p 00000000
00:00 0 173517492 173513320 173513320 173513320 173513320 0
0 [heap] 2b2fa8000000 rw-p 00000000 00:00 0 11026824
11007512 11007512 11007512 11007512 0 0 2b2f62000000 rw-p
00000000 00:00 0 1146880 1014280 1014280 1014280
1014280 0 0 2ae9072b0000 rwxp 00000000 00:00 0
71808 69280 69280 69280 69280 0 0 2ae917d88000
rw-p 00000000 00:00 0 25696 24700 24700 24700
24700 0 0 2b325732f000 rw-p 00000000 00:00 0 16384
16328 16328 16328 16328 0 0 2ae91aeb9000 rw-p
00000000 00:00 0 9988 8972 8972 8972
8972 0 0 2b3249063000 rw-p 00000000 00:00 0 9216
8204 8204 8204 8204 0 0 2ae905332000 r-xp
00000000 08:02 8391747 13168 6808 2825 6804
0 0 0 libjvm.so*
* additional information:*
*hadoop version: 2.7.2*
* -Xms280G *
*-Xmx280G *
*-XX:MaxDirectMemorySize=10G *
*-XX:MetaspaceSize=128M*
*-XX:+UseG1GC*
any idea will be appreciated
