| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 761.1 | Cache Policy and selecting NOWRITEBACK | SSDEVO::THOMPSON | Paul Thompson, Colorado Springs | Fri Feb 07 1997 11:01 | 41 |
| See answers by the >>>
Hi!
Again one questions about RAIDsets when the battery fails on
HSZ40:
1. Question:
Is there any difference beetween a UNIT (RAID or Mirrorset)
that have Writeback-Cache enabled a UNIT (RAID or Mirror)
which is set to NOWRITEBACK when a Battery fails.
>>> Even when NOWRITEBACK is selected, RAID and Mirror sets require that WBC
>>> be operable. If WBC is inoperable because the battery has failed, RAID
>>> and Mirror sets are inaccessible through that controller.
As far as i understand now with HSOF 2.7 both UNITS will
go INOPERATIVE. Is it possible that one UNIT stays online
a little longer than the other?
>>> See above answer.
2. Question:
How does the CACHE_POLICY take effekt with V3.0?
I can not imagine that there will be any difference
if CACHE_POLICY is A or B, in both cases the unit will failover to the
other (good) controller or not?
Is there a recomended setting for CACHE_POLICY in dual-redundant
configurations with HSOF 3.0?
>>> No changes to cache policy took place in 3.0. Cache policy has no impact
>>> on failover.
>>> If you want to be certain that RAID and Mirror sets will only work when
>>> the battery is fully charged, select cache policy "A".
HELMUT
>>> Paul
|
| 761.2 | Thanks! (explanation) | ATZIS2::PUTZENLECHNE | wherever is fun, there's always ALPHA | Fri Feb 14 1997 07:52 | 10 |
| Thank You for your explanations.
I do aggree that CACHE_POLICY has no effect for failover, but my
opinion is that with V3.0 it makes no sense definiung a CACHE_POLICY for
dual-redundant controllers, since in every case of a battery-failover
the controller will shutdown and units will failover to the other
controller regardless of the CACHE_POLICY-setting. That is why with
V3.0 CACHE_POLICY only is usefull for single-controllers.
Helmut
|
| 761.3 | Not quite true | SSDEVO::RMCLEAN | | Fri Feb 14 1997 09:04 | 5 |
| Not strictly true. Should you replace batteries on both controllers there is
some risk that neither controller will have fully charged batteries. Cache
Policy B permits 10 hours of access to the drives before it declares a problem.
This gives you time to charge the batteries. Since with the HSZ50 you have
both batteries as a FRU there is a lot of use of a different policy.
|
| 761.4 | | OTOOA::JPOND | | Wed Mar 26 1997 08:13 | 11 |
| Could you elaborate on this statement in -.1:
> Since with the HSZ50 you have
> both batteries as a FRU there is a lot of use of a different policy.
How does the HSZ50 act, with its different cache and battery
architectures (and its different firmware), in relation to how the
HSZ40 acts, when it comes to battery failures and the setting of cache
policy?
Jim
|
| 761.5 | The code is identical the batteries are packaged differently | SSDEVO::RMCLEAN | | Wed Mar 26 1997 08:34 | 1 |
| The batteries are different the code is exactly the same.
|