topic Re: Why Cache Batteries Fail in Disk Enclosures

Why Cache Batteries Fail

Khurram Shahzad — Mon, 29 Jun 2009 07:41:52 GMT

Hi All,

Can you any one tells me the reason why Cache batteries fail in EVA. I mean is it due to improper shutdown or some internal battery component failure?

Re: Why Cache Batteries Fail

Víctor Cespón — Mon, 29 Jun 2009 08:22:47 GMT

The batteries are composed of several elements, usually 3.
The controllers perform a periodic charge and discharge of each one of the elements, to test if the battery has enough charge to mantain the information on the cache for 96 hours.

If some element does not has enough charge, the battery is declared "low charge" and attempts to charge it. If it does not charge, is declared "failed".

Re: Why Cache Batteries Fail

Khurram Shahzad — Mon, 29 Jun 2009 08:27:24 GMT

Thanks for your reply. Can you please tell me any phenomena or reason that cache batteries fail due to improper shutdown of EVA.

As i have heard this can be one of the reason is it true?

Re: Why Cache Batteries Fail

Víctor Cespón — Mon, 29 Jun 2009 09:21:10 GMT

There's no direct relation. I've seen hundreds of EVA logs, and in some of them there are unexpected reboots of the controllers due to power failures, but they are not followed by a battery failure.

It can happen on some case, like any other component. I've had cases of a disk or a fan failing after powering the EVA back on. It just means that that component was in bad health and did not take well the power-cycle.

Re: Why Cache Batteries Fail

Khurram Shahzad — Mon, 29 Jun 2009 09:36:46 GMT

Bundle of Thanks

Re: Why Cache Batteries Fail

Pieter 't Hart — Tue, 30 Jun 2009 07:00:27 GMT

In the case that the EVA is shutdown and powered-down for a considerable time, the batteries still try to apply power to the cache-memory.
you may need an extra action to "shut down" the batteries, so they dont fully discharge.

Re: Why Cache Batteries Fail

Greybeard — Tue, 30 Jun 2009 08:19:47 GMT

is this a one off or have your batteries failed more than once? There was a bug in one of the xcs code rev's that meant the controller had to be powered off (the norm is to hot swap) for the replacement, or the battery could be tagged as bad but the failure would not occour till the next test cycle which I think is about 90 days later. It was rev 6. something I believe.