Backout issues with upgrading firmware

[John A. Beard]

Hi folks,

I would appreciate your advice on the subject of upgrading the firmware on a pair of EVA4000 controllers (located at two different sites).

We are about to have the firmware on our EVA400 controllers upgraded from XCS 5.031 to XCS 6.220. Our immediate concern relates to how we can possibly recover should the upgrade fail (Murphy's law). We have been led to believe that all the SAN based information would be lost and that we would have no option other than to recreate the environment from scratch and having to restore all the user data from backup media.

I am aware of how to use SSSU to capture the information at both sites, and also that we would have the locally attached SCSI drives mirror the contents of our SAN system disk in order that we can boot all the 8 VMS servers that would be impacted. The latter would enable us to communicate with the ADSM backup server for the purpose of restoring data.

In a nut shell, does anyone have knowledge as to how this process could be carried out so the original (existing) controller information would be secure and re-usable should the upgrade fail (is SSSU capture all we need?)

I apologize if I have omitted other bits and pieces that might allow you to answer this question. I have only a small amount of exposure to the SAN world and would appreciate any help that you might be able to offer.

FYI The data is replicated at both locations via VMS host based volume shadowing.

[AnthonySN]

upgrade direclty to xcs 6.220 is unsupported
it has to be v5.100, v5.110 or v6.000
and there is a time line you need to wait for each firmware level upgrade.( i think 3-4 hrs for each level)
my suggestion would be to upgrade the dr site first.
also stop replication when doing the primary eva upgrade.

have a good backup of the data to tapes.

[AnthonySN]

forgot to mention
1. 5031 to 5032 (5032 can get from HP)
2. 5032 to 5110
3. 5110 to 6110
4. 6110 to 6200

special firmware 5.032 requires downtime, you can apply it and from this version the upgrades are online without the downtime.

the 5110 firmware is here:
ftp://ftp.hp.com/pub/softlib/software7/COL15054/co-40158-1/HSV200_5110.zip

the 6100 firmware here:
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc
=us&prodNameId=470478&prodTypeId=12169&prodSeriesId=470490&swLang=13&
;taskId=135&swEnvOID=54


the 6200 firmware here:
http://h20293.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=
T4256-63113A

you need to upgrade the HD firmware also
which is below http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc
=us&prodNameId=3453487&prodTypeId=329290&prodSeriesId=3302267&swLang=
13&taskId=135&swEnvOID=54 "

also you need to upgarde the Command view EVA.
CV-EVA 7 onwards is licensed.
also you need to check the hba firmware revisions of the servers also.

[John A. Beard]

Hi,

Thank you for taking the time to respond.

To clear up some of your points...

Command View is at 8.0, HBA firware is current and we recently applied the latest patches to VMS (including volume shadowing).

To prevent shadowing of data we would normally disolve the VMS shadow set in cases where wanted one EVA site to remain up. In this case (because of time constraints) we need to upgrade both sites simultaneously, so we don't have a grace period where we could revert back to using only one site's EVA is the other failed.

In the case of a controller upgrade (we are aware of the different stages), the original plan was to shut down both controllers at each of the two sites and perfrom the upgrade to each set in parallell (different staff at each site performing the upgrade).

Som apart from a SSSU configuration captute, is there any way of protecting the disks themselves during the whole process..ie, even if the data was still there , would it be inaccessible should the firmware upgrad go belly up.

[Uwe Zessin · Accepted solution]

The problem is not that the firmware flash could fail - if nothing helps one can swap the firmware cards in the controller. The real (potential) problem is that almost every software upgrade of VCS/XCS/CV-EVA makes irreversible meta-data changes.

[John A. Beard]

Thanks.. the reference to the change in meta-data sums up what I was looking for. We will make the necessary arrangements to have all our image backups in place prior to proceeding...because of the nature of the data we can spread this ove a the week before.

[John A. Beard]

Thanks for all the information