Synergy Profiles and WWNs

[DWOOH]

Backstory:  We bought a bunch of Synergy 480 Gen10 with SD card boot for VMware hosts shorty before VMware admitted their 7.x re-design put the nail in the coffin for SD boot.  So now we have to replace SD with SSD before we upgrade to ESXi 7.x

This requires a Synergy Profile change to identify the local disk as the boot device rather than the SD card.  We will then need to install the new OS and put the blade back in service.

We use an external, fiber attached storage array.

So the question is this:   If I create a new Profile Template with the SSD as boot device, and apply it to the blade, will the blade get a new set of WWNs?

I'd rather it didn't so once the blade is rebuilt I can just add it back to the cluster and it will have access to the storage without needing to rezone.

Other options include:

Modify the host profile, and live with the "Out of compliance" warnings

Modify the Profile Template and live with Out of compliance warnings until complete.  But I would be concerned an inadvertent reboot before SSD is installed would render the blade useless until addressed.

Thoughts?   Suggestions?

And .... thanks VMware  >8^(

[PatrickLong · Accepted solution]

Yes, if you create new profiles then you will get new WWN's and MAC's. There are ways to manually apply the old values to the new profile, see @ChrisLynch  comment here:  https://community.hpe.com/t5/HPE-OneView/Apply-new-profile-server-for-active-server/m-p/7161588/highlight/true#M6960

Other options:

1) If you require that OneView manages the boot order, change the consistency checking to Not checked and then Reapply the template to your profiles. I *think* you can do this with the compute modules powered on because you are only unmanaging a setting and not the reverse, but I have not actually tried it with this Manage Boot device setting; I HAVE tried it with the Firmware baseline setting and it is allowed to change to unmanaged with host powered on. Then as you install the new boot drives on each compute module and AFTER (or as part of) your process to relocate/recreate the boot device, change the boot order on that server via RBSU in iLo console at boot time - PITA, I know. but you will not have Compliance warnings because you are not checking that setting for Exact Match.  Lather, rinse, repeat. When all hosts with profiles based on that template have been remediated, go back to OneView and edit the profile template to set the Primary boot device to Hard Disk and set Consistency checking back to Exact Match and Refresh your profiles from template.  This part will likely present you with the same out of compliance issue you suggested because you're trying to apply new management settings to a powered-on host. I believe another reboot would be required to bring it into compliance.

OR

2) Simply uncheck the box for Manage Boot order in OneView altogether and manage your boot device settings via RBSU manual changes.  Honestly how many times are you gonna need to change that setting anyway - likely just this once to accomodate the relocation of the boot device I would think?  Do you really need OneView to manage it on every boot cycle?

If you insist on retaining existing profiles I don't see how you get through it without some period of "Out of compliance" warnings you noted.

And +1 to your VMware comment. they DEFINITELY underestimated the impact of changing the boot device partitioning, filesystem changes to VMFS-L, and boot device I/O profile changes on diskless ESXi servers, and very likely underestimated the pervasiveness of this type of configuration among their install base. And then compounded the problem by not giving enough fair warning of the change to allow customer multi-year procurement cycles to address it in advance.  I'm exclusively diskless in my c7000 and Synergy environments and have a LOT of these remediations in my future.  For now I'm stable on ESXi 7.03 U3c - soon to be U3d - from the latest HPE custom image profile and Latest 2022.02.01 SSP firmware. (+newer BIOS, which is allowed.)

[DWOOH]

Thank you!

Yes I think the "Unmanaged" option would be the way to go, and hit it from the iLO when I'm ready to re-build.

We were slow to the table for SD boot. Tried it when we got our first Synergy frames on a few blades.  Thought "This is Great!"  So our next big refresh cycle of +50 blades was all SD .. then came ESXi 7...

I actually installed 7.x and felt the pain.  Even with all their special settings, hosts would still lock up and crash.  I went back and re-installed 6.7    Now I'm "ready" to re-reinstall 7 ....

Thanks Again!

[ChrisLynch]

The thing to remember here is that a server profile represents the personality and expected configuration of a server.  In this case, managed connections by default with Virtual Connect will manage the hardware address.  In order to ensure that duplicate addresses are not used, new addresses are assigned when you create profiles.

You can override this by specifying a user defined address in the connection.  Do know that you can also override if address assignment should be Virtual or Physical.  Once you create the profile, you cannot change that policy.

[Chris_Coates]

Newer servers I buy have SSD for boot, but the ones that we bought before ESXi7, that are booting off SD, were upgraded to 7.03c and are running fine. It is no longer recomended but still supported. I worry that before the hardware is EOL it might not be supported, but for now it is and is working fine in both ESXi and VSAN.

[DWOOH]

We're now embarking on this adventure....

So I updated the Server Profile Template to Boot Settings - Boot Device = Umanaged, as well as SSP update.  And as expected the blades using that profile flagged Inconsistent

The blade is removed, ssd installed, and plugged back in, and it cycles through and inventories the blade.

For the Server Profile, I "Update from Template" and it applies the SPP.   When that is done, I select "ReApply" from the "Actions" menu.  From what I understand, either of those should update the Server Profile settings according to the attached Server Profile Template.

However, the Boot Settings is still set to "SD card", and there are no "Inconsistent" warnings.  What am I missing?

The server still boots to the SSD and ESXi 7 is installed.  And even with a power off and power on, it still boots and works fine. - and YES it did keep the WWNs and attached back up to the storage array.

Just something unexpected with the profiles.

[DWOOH]

It hit me last night - "Unmanaged" setting in the Profile Template apparently does not to check the setting in the profile.  An "Update from Template" does not change the profile to Unmanaged, it just leaves it as it is currently set.

I updated the Server Profile with "Unmanaged" for boot device, "Refresh", and the profile still reports "Green" and "Consistent"

Analogous to an Active Directory GPO setting of "Not Set"

[ChrisLynch]

In the "Server profile is inconsistent with associated profile template" warning message, if you expand the Details link you will see exactly what is not consistent.  It will also tell you if the inconsistency can be remediated online or require manual changes.