PowerFlex: How to Replace NVDIMM in 15G and later PowerEdge Node

Before you begin:

Save the output of the following commands in a notepad:

• Log in to the Primary MDM with admin login. use single quotes ' ' if password has special characters

scli --login --username admin --password '<password>' --management_system_ip <pfmp_ip>

• Run the following command to list all SDSs details:

scli --query_all_sds

Example:

SDS ID: 2e400d3800000004 Name: LAB10_SDS3 State: Connected, Joined IP: XXX
SDS ID: 2e400d3700000003 Name: LAB10_SDS2 State: Connected, Joined IP: XXX
SDS ID: 2e400d3600000002 Name: LAB10_SDS4 State: Connected, Joined IP: XXX
SDS ID: 2e400d3500000001 Name: LAB10_SDS1 State: Connected, Joined IP: XXX
SDS ID: 2e400d3400000000 Name: LAB10_SDS5 State: Connected, Joined IP: XXX

• Location the impacted SDS and note down the SDS name then run the following command, output will help later in step 10 to put back the devices using device path

 scli --query_sds --sds_name <sds_name>

• Finally run the following command from the impacted SDS:

dmidecode --type memory | grep "Non-" -B 3 -A 3 | grep -E 'Locator|Serial' | grep -v Bank 

ndctl list -D 

ndctl list -N 

ndctl list -R

Procedure

1. login to PFMP > Block > SDSs
2. Select SDS in question and click on devices on the right pan

3. Remove all the Storage Devices from the SDS and wait for the Rebalance to finish. (Don't remove the Acceleration Devices in this step)

4. Once all the devices are removed from the cluster, remove all the Acceleration Devices (ex: /dev/dax0.0)
5. Once the server is no longer contributing any storage, power off the server and let the CE replace the NVDIMM. 
   
   Caution: Don't move to Step 6 or Sanitize NVDIMMs until all SDS devices are removed
6. Power on the server and Boot into the BIOS. Select BIOS Settings > Memory Settings > Select Enabled for Sanitize All NVDIMMs.

See PowerFlex: How to Sanitize NVDIMM and SDPM - Video

7. Save the changes and exit, then boot the server.
8. Run the commands:

   ndctl create-namespace --region=region0 --type=pmem --mode=devdax --align=4k --force

   ndctl create-namespace --region=region1 --type=pmem --mode=devdax --align=4k --force 

   Example:

ndctl create-namespace --region=region0 --type=pmem --mode=devdax --align=4k --force
{
 "dev":"namespace0.0",
 "mode":"devdax",
 "map":"dev",
 "size":"31.50 GiB (33.82 GB)",
 "uuid":"44490109-0bff-4ca2-b68a-ae7a6b30a55f",
 "daxregion":{
 "id":1,
 "size":"31.50 GiB (33.82 GB)",
 "align":4096,
 "devices":[
 {
 "chardev":"dax0.0",
 "size":"31.50 GiB (33.82 GB)"
 }
 ]
 },
 "align":4096
}

ndctl create-namespace --region=region1 --type=pmem --mode=devdax --align=4k --force
{
 "dev":"namespace1.0",
 "mode":"devdax",
 "map":"dev",
 "size":"31.50 GiB (33.82 GB)",
 "uuid":"d48959c9-4b8c-4736-ad00-23afc0113fcd",
 "daxregion":{
 "id":1,
 "size":"31.50 GiB (33.82 GB)",
 "align":4096,
 "devices":[
 {
 "chardev":"dax1.0",
 "size":"31.50 GiB (33.82 GB)"
 }
 ]
 },
 "align":4096
}

Each DAX device name is displayed in the output as a chardev value. 
In the example output, DAX device names are dax0.0 and dax1.0. 
The full acceleration device names are /dev/dax0.0 and /dev/dax1.0

9. Go back to PFMP > SDS > Select the SDS > Add Device > Acceleration Device (ex: /dev/dax0.0 and /dev/dax1.0) (add the acceleration device first before Storage Device) 
10. After adding the Acceleration Devices, you can add back all the Storage Devices noted in step 1 (ex: /dev/XXX or if using cloudlink /dev/mapper/svm_XXX) , you can refer back to the output of the command  scli --query_sds --sds_name <sds_name> that was saved earlier 

For replacement using PFMP wizard see PowerFlex 4.X: How to Replace NVDIMM using PFMP Wizard

Related Articles:

• PowerFlex: How to correlate between NVDIMMs DAX and SDS devices
• PowerFlex: How to check NVDIMM Health Life Used Percentage
• PowerFlex: How to Sanitize NVDIMM and SDPM - Video
• PowerFlex: Sanitize NVDIMM after NVDIMM firmware downgrade