EVA 4400 - Dual battery fail - still online ??

[larsjohansen5]

Hi,

I was at a new customer today to go through his EVA set up.

It was set up three years ago, and nothing had been done to it since that day.

Both Battery modules on the controller were yellow, and in CV they come up as failed, still the EVA was online.

From what I have heard, the EVA will take the virtual disks offline when both batteries are dead? Have I been misleaded or is it due to that they run very old firmware 0900400

[Johan Guldmyr]

Hi,

when both batteries are broken then the vdisks should not be visible from the hosts.

It is likely that the batteries haven't actually failed due to something lovely in that old firmware. Or that they've failed and nothing's happened.
Is the cache policy on write-through on the vdisks?

I'd presume that the controller event logs should be able to tell more about what's going on.

However, between 09522000 and 09501x00 there aren't mentioned any battery fixes (09522000 release notes). Nothing specifically relating to the same in 09534000 or 10 either. But maybe it's not public.

[larsjohansen5]

Hi,

The volumes are in write-back mode, so its kind of strange.

My first though was to replace the batteries, hopefully they show up green, and then do a firmware upgrade.

If the problem is firmware related and reseating them dont work, can I do the upgrade of firmware if both batteries show up as dead in CV? I though you could not update when you have yellow lamps 

[Johan Guldmyr]

If possible, it would be nice to get some input from the support after they've gone through the logs.

You can upgrade when you have yellow leds, the idea behind not doing it would be that the array is not in a stable state. Generally you do want to do firmware upgrades when it's stable. But sometimes you do not have a choice.

Something to investigate: let's say both batteries are dead after reseat of batteries or controller restarts and hosts cannot see the LUNs. Does that mean your CV server also cannot manage the EVA?

[larsjohansen5]

This EVA does not have a support contract anymore, we will try to get the customer to register a new one.

Reseating the battery module did not help so we are going to order a new one, to see if that fixes it.

[Johan Guldmyr]

Maybe controller restart if you dare.

[larsjohansen5]

Ok,

I talked with a HP technician in private about this, he told that it is a misunderstanding that the LUNS go offline, they just disable write back.

Im going to upgrade from 09004000 to 09534000(says it supported in the reference guide)

In that regard, what do I do with Command View, according to the  compatability guide these two firmware version does not have any CV version that works for both, do I upgrade with the version I got now(8.x) and when the firmware install is finished, I upgrade/reinstall to 9.4 or 10?

[PDP-Fan]

It seems there is a lot of confusion around these Batteries. I have here an EVA4000 and 5000 and both refuse to present vdisks when all batteries are dead.

The batteries of the 5000 are easy to repair because the module has not many active electronic devices. So I was able to make that EVA work again. 4000 batteries are more complicated.... i guess i will have to buy new ones to make my 4000 present it's vdisks.

(PS: these are my test systems in the lab, no productive jobs run on them)

[PDP-Fan]

Addendum:

It does not matter if cache policy is write-back or write-through or if read cache is on or off.... dead batteries mean unpresented vdisks. I just tested that...