HPE Storage Users Group

Having recently upgraded to 3.1.2(MU3) I've read quite a lot about persona changes on ESXi hosts etc

Has anyone had any experience of this ?
What is actually required on both the Array and the VMHost itself ?


Thanks
Dave

VM Host has to be powered off to make the change. The new persona I believe is required in a future update.

I believe 3.1.3 is the last release to have the old VMware persona, the new one is the better one to use moving forward using ALUA, and is required for things like peer persistence.

Can this be one just one host at a time?

I haven't done this with VMWare as our array started on 3.1.2 so we started with the new persona, but I did just yesterday change some Windows boxes inadvertently. I was in double checking everything for our upcoming hardware expansion and OS upgrade and I noted that some of our online imported hosts that were auto-created during the migration process weren't showing the full OS version, and when I changed them it also changed them from persona 1 to persona 2 without any ill effect.

Richard Siemers wrote:Can this be one just one host at a time?

I would have thought so, even with hosts in a host set, as I've seen VMware servers and a windows server in one before I think.

From a VMware cluster pov it would be fine doing it one at a time.

One thing to note though is if you had changed the default satp setting on the host via the cli, as the default behaviour has changed from active/active to ALUA.

With host persona 11 it would be
# 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 "HP 3PAR Custom iSCSI/FC/FCoE ALUA Rule"

With persona 6 it would have been
# esxcli storage nmp satp rule add -s "VMW_SATP_DEFAULT_AA" -P "VMW_PSP_RR" -O iops=100 -c "tpgs_off" -V "3PARdata" -M "VV" -e "HP 3PAR Custom iSCSI/FC/FCoE ALUA Rule"


Also remember that the latest best practice also states that you should change the iops from 1000 to 1 (it used to be 100)

As painful as it is; just get it done.
Might as well do it as you implement the new claim rules with IOPS=1 anyway...

Also remember that the latest best practice also states that you should change the iops from 1000 to 1 (it used to be 100)

Where did you get this information from ?

It is in the updated VMware best practice guide. The ESX implementation guide still says IOPS=100, but HP has confirmed IOPS=1 is the new best practice.

IT might be worth my while trying to get HP to confirm this for me too ... seeing as we have a bunch of persona changes to do!