[CentOS] Unbound nodes using numactl

Mahmoud A. A. Ibrahim

m.ibrahim at compchem.net
Fri Jul 24 22:48:41 UTC 2015


Dear Centos Users
We were wondering how can we change the default policy of numactl.
In other words, our cluster has 10 servers, each server has two nodes, and
each node has 16 cores.
Recently, we observed a poor performance of our cluster when a couple of
jobs run on the same server. By investigating the problem, we found that
each job moves randomly among the 32 cores of the server. This returns to
the fact that the 32 cores of each server bound together as seen below.
How can we make our server back to two numa nodes?
Any kind of support will be highly appreciated.
B.T.W. We use Rocks6.1
Thanks in advance
Sincerely;
M. Ibrahim

################
#numactl -show
policy: default
preferred node: current
physcpubind: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
24 25 26 27 28 29 30 31
cpubind: 0 1
nodebind: 0 1
membind: 0 1
################
# numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
node 0 size: 16287 MB
node 0 free: 15619 MB
node 1 cpus: 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31
node 1 size: 16384 MB
node 1 free: 15839 MB
node distances:
node   0   1
  0:  10  21
  1:  21  10
################



More information about the CentOS mailing list