Controller battery failure

[viks]

Hi All,i got an AR12H with two controllers.sometimeback i changed the second battery on X controller.but after a few days it showed bad.so again i replaced the both the batteries of the X controller with a good set.again the same battery #2 is failing.if started on monday till wednesday there is no problem but later "replace x battery" is displayed.i did a small preventive maintenance also.but no use.how to approach this.any help wud be highly helpful.
thanks in advance

[Bill McNAMARA_1]

I tried to avoid this question, but seeing as nobody has replied, will give you what I believe is bad news: Controller Failure... and a very stupid type of failure..

Again change the 2 batteries, to known good, fully charged ones, and see if Battery 2 still fails. If so, get the ouptut of the logprint command for the arrayid and send it to your hp rc engineer.

There is a risk of data failure and performance is going to be poor because of disabled caching.

The data failure will come if there is a power fail or incorrect shutdown and you are using a strange controller map resiliancy.

Post up your arraydsp -a of the array.

Good luck,
Bill

[viks]

Dear Bill ... sorry for the extreme delay.thanks for ur response.but again i was observing this week.two days continuously it works fine but the third day it tells battery failure.i got one more question.can i just swap the X and Y controllers so that i will be sure that my system doesnt goes down at any point of time.now i am not very sure if X fails i Y will take care or not.theoritically it shud.but if it fails then i am in a soup.right.reply me at the earliest.
regards
Viks

[Bill McNAMARA_1]

Hi again,

I'll look into your attachment in a minute, but just to reply to the failover case of the controller first..

The system should not go down on any failure in the autoraid, unless you have your boot disk/swap fs there.

BUT, certain filesystems on the autoraid are potentially at risk if you haven't configured LVM to respond correctly to "failed hardware"

ie,
Lets imaging you have the following configuration

Controller X SCSI ID 1
Controller Y SCSI ID 2

LUN0 LUN1

Controller X connected to host directly.

Controller Y connected to host directly (not the same bus as Controller X)

See diagram attached

Make sure that LVM has been configured for switchover, via vgdisplay All_VGs_on_12H and look for Alternate Paths to the luns, if you don't have an alternate path defined, there will be no switch over and your filesystems /vg on the 12H will go down brutally.

Bill

[Bill McNAMARA_1]

1 battery in controller X has for sure failed:
Battery #1 state = DEAD
Battery #2 is okay.

I'd swap position of the two batteried around just to verify that it is either the controller that is damaged or if it is the battery itself.
Run the same arraydsp command.

If you see
Battery #1 state = DEAD
Battery #2 state = GOOD

Then your controller has failed, or if you see

Battery #1 state = GOOD
Battery #2 state = DEAD

after swopping the X controller battery positions around then the Battery (and not the controller) is a dud.

I can't exactly recall which is battery #1 and #2, but recommend to change BOTH at the same time, (don't want a weaker battery that the other)

Try to get a thin pliers to remove/intert the battery connector, because you can a> hurt your fingers! and b> pull out wires from the battery connector.

On the brighter side, your array looks in good health,
(There is f/w upgrade available, however, I wouln't recomment upgrading just for the sake of it.)

Good luck,
Later,
Bill

[harry d brown jr]

viks,

Plus,read this:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x96b70bce6f33d6118fff0090279cd0f9,00.html

live free or die
harry

[Bill McNAMARA_1]

I guess point are out of the question for my lovely art! ;P

Later,
Bill

[viks]

Dear Bill , tell me onething.i am sure my raid box wont go down as all the vg's have got alternate paths.but what i want to avoid is failure of X controller because of battery dead condition.so i thought i can interchange X and Y controller positions and run the show for sometime till i get a replacement controller.can i do this.
regards
Viks

[Bill McNAMARA_1]

Hi Viks,

The autoraid will never perform a true non-interupted failover. you will get application hangs, for the duration of the pvchange switchover, should be set to around 90secs for the autoraid.

However, if your applications behave well, pulling a controller will not cause a problem.

Changing the X and Y physical position will make no difference to the operation of the autoraid. Being in either slot X or Y has no bearing on the functionality of either controller.

You can pull the X controller and the array will run fine off the Y controller.. without moving it anywhere.

It will work fine in slot Y, once you have verified that the alternate links are good.
ie the Y controller will be used on X removal. There is nothing more that you need to do other than removing the failed controller.

(At least test that once you've identified it is certainly not a battery failure by swopping battery positions in the controller)

Later,
Bill