VMware HPE 3Par recommended settings

Made the following script yesterday to run on all our ESXi hosts to apply HPE recommended settings for 3Par.

#!/bin/sh
# Set VMWare recommended path settings for our datastores
# https://h20195.www2.hpe.com/V2/getpdf.aspx/4AA4-3286ENW.pdf
# Note: this only targets devices of identifier naa.6000xxxx. Modify this later if required.

for i in `esxcli storage nmp device list | grep '^naa.6000'` ; do
        echo "Setting RR for $i"
        esxcli storage nmp device set --device $i --psp VMW_PSP_RR;
        echo "Setting IOPS=1 for $i"
        esxcli storage nmp psp roundrobin deviceconfig set --type=iops --iops=1 --device $i;
        echo "Setting queue size for $i"
        esxcli storage core device set --device $i --queue-full-threshold 4 --queue-full-sample-size 32;
        echo ""
done

echo "Setting as default for future"
esxcli storage nmp satp rule add -s "VMW_SATP_ALUA" -P "VMW_PSP_RR" -O "iops=1" -c "tpgs_on" -V "3PARData" -M "VV" -e "HPE 3PAR Custom Rule"

#!/bin/sh

# Set VMWare recommended path settings for our datastores

# https://h20195.www2.hpe.com/V2/getpdf.aspx/4AA4-3286ENW.pdf

# Note: this only targets devices of identifier naa.6000xxxx. Modify this later if required.

for i in `esxcli storage nmp device list | grep '^naa.6000'` ; do

echo "Setting RR for $i"

esxcli storage nmp device set --device $i --psp VMW_PSP_RR;

echo "Setting IOPS=1 for $i"

esxcli storage nmp psp roundrobin deviceconfig set --type=iops --iops=1 --device $i;

echo "Setting queue size for $i"

esxcli storage core device set --device $i --queue-full-threshold 4 --queue-full-sample-size 32;

echo ""

done

echo "Setting as default for future"

esxcli storage nmp satp rule add -s "VMW_SATP_ALUA" -P "VMW_PSP_RR" -O "iops=1" -c "tpgs_on" -V "3PARData" -M "VV" -e "HPE 3PAR Custom Rule"

No original contribution from my side here. It’s just something I put together from stuff found elsewhere.

I wanted to run this as a cron job on ESXi periodically but apparently that’s not straightforward. (Wanted to run it as a cron job because I am not sure the queue settings too can be set as a default for new datastores). ESXi doesn’t keep cron jobs over reboots so you to modify some other script to inject a new crontab each time the host reboots. I was too lazy to do that.

Another plan was to try and run this via PowerCLI as I had to do this in a whole bunch of hosts. I was too lazy to do that either and PowerCLI seems a kludgy way to run esxcli commands. Finally I resorted to plink (SSH was already enabled on all the hosts) to run this en-masse:

Get-Content .\esxlist.txt | %{ $entry = $_ -split ": " ; $esxhost = $entry[0]; $pass = $entry[1]; "Processing $esxhost";  .\plink.exe -pw $pass root@$esxhost /vmfs/volumes/path/to/fixPSP.sh }

1	Get-Content .\esxlist.txt \| %{ $entry = $_ -split ": " ; $esxhost = $entry[0]; $pass = $entry[1]; "Processing $esxhost"; .\plink.exe -pw $pass root@$esxhost /vmfs/volumes/path/to/fixPSP.sh }

This feels like cheating, I know. This requires SSH to be enabled on all hosts. This assumes I have put the script in some common datastore accessible across all hosts. I am using PowerShell purely to loop and read the contents of a text file consisting of “hostname: password” entries. And I am using plink to connect to each host and run the script. (I love plink for this kind of stuff. It’s super cool!) It feels like a hotch-potch of so many different things and not very elegant but lazy. (Something like this would be elegant. Using PowerCLI properly; not just as a wrapper to run esxcli commands. But I couldn’t figure out the equivalent commands for my case. I was using FC rather the SCSI).