Table of Contents
Advertisement
4.
Confirm the memory size the OS recognized before setup of memory mirroring.
You can get the memory size of each NUMA node by using the following numactl -H.
Execution resultEx
# numactl -H
available: 4 nodes (0-3)
node 0 cpus: 0 4
node 0 size: 16208 MB
node 0 free: 15600 MB
node 1 cpus: 1 5
node 1 size: 16384 MB
node 1 free: 15936 MB
node 2 cpus: 2 6
node 2 size: 16384 MB
node 2 free: 15991 MB
node 3 cpus: 3 7
node 3 size: 16384 MB
node 3 free: 15957 MB
node distances:
node
0
1
2
0: 10 15 15 15
1: 15 10 15 15
2: 15 15 10 15
3: 15 15 15 10
5.
Enable memory mirroring.
Execute necmemras with "--mode" and "--mirrornode" options for enabling memory mirroring,
and reboot the OS to apply the settings.
Execution resultEx ( The mirrored NUMA nodes are 0 and 2.)
# ./necmemras --mode=independent-mirror --mirrornode=0,2
Memory RAS Mode: Independent + Mirroring
Node:0 MR#1-2 Mirrored (Base Memory)
Node:1 MR#3-4 Not-Mirrored
Node:2 MR#5-6 Mirrored
Node:3 MR#7-8 Not-Mirrored
The configuration changes have not been applied yet.
You must reboot the system to apply them.
6.
Check to verify that the memory configuration actually applied.
The memory size of the mirrored NUMA node is reduced to about to half size. After system
reboot, check that the memory size of the mirrored NUMA node confirmed by step 4.
If the memory size is not reduced, the memory component of the NUMA node may not satisfy
the populating rule for memory mirroring. In this case, check the activity log to see the reason.
You can check the activity log by [Maintenance] -> [Hardware Log] -> [Display Activity Log] on
Web console. Refer to "Express5800/A1040d, A2040d, A2020d, A2010d User's Guide" for
detailed information.
3
4
Advertisement
Table of Contents
