https://www.jianshu.com/p/074c43e1f9c6
$yum install -y numactl
[root@localhost ~]# numastat
node0
numa_hit 30772351
numa_miss 0
numa_foreign 0
interleave_hit 13741
local_node 30772351
总结:
1、num_miss是比较关键的参数,如果此值过高的话,说明当个node的内存命中率比较低,此时可能需要绑定cpu战略
2、使用参数-p可以查看单个进程在不同node下,内存的分布情况
[root@localhost ~]# numastat -p 847
Per-node process memory usage (in MBs) for PID 847 (vmtoolsd)
Node 0 Total
--------------- ---------------
Huge 0.00 0.00
Heap 0.59 0.59
Stack 0.04 0.04
Private 5.79 5.79
---------------- --------------- ---------------
Total 6.42 6.42
3、使用监控命令watch -n1 --differences numastatj监控miss值变化,如果太高的时候可以考虑绑定node
[root@localhost ~]# watch -n1 --differences numastat
numactl --hardware
[root@localhost ~]# numactl --hardware
available: 1 nodes (0)
node 0 cpus: 0
node 0 size: 2047 MB
node 0 free: 214 MB
node distances:
node 0
0: 10
[root@localhost ~]# numactl --show
policy: default
preferred node: current
physcpubind: 0
cpubind: 0
nodebind: 0
membind: 0