| The design requirements/goal was that the battery be able to maintain
memory for a minimum of 100 hours. That is a very conservative number
and there is evidence that it would be maintained 2x or more of that
figure with 2 fully charged batteries.
If the power off condition exceeds that amount of time, then its likely that
the customer has more pressing problems at that point than the loss of
data in cache......like his/her home refrigerator at home has defrosted AND
the significant other is calling!!
You are correct, with cache_policy=A, at startup time, you will not be able
to bring a raidset up until good battery status is attained.
There is consistency information in written in cache to ensure that what
is in cache is still good data so that when things are operational, that
the data will be appropriately delivered to the media and be accurate.
They are trying to protect the data!
|
| Soory for later feedback!
The situation for 100 hours could be accepted. But we happen to have 5
hsj42 (10 *hsj40) controller lost power suddently for 2 hours.Then ups
charge for 1 hours , lost power again for 3 hours. But when power back,
only one hsj40 battery charge good in 15 minutes.The other all display
with batterly low. then waiting 12 hours later, some of the controller
becomes bad battery. The battery rev is "B01", some batteries without
rev labels though they are new. Could this be battery production
problem? I had thease events for 3 times. For such situations, what
action should we done ?
Thanks,
Fred.
|