IBM System Storage DS4000, Руководство по оборудованию

Страница: 432 / 566

414
IBM Midrange System Storage Hardware Guide
Clearing all error counters
We have already mentioned that there are options in Storage Manager to reset RLSD
baseline and clear drive channel error counters. A neater way is to reset both of these and the
SOC statistics with a simple script that can be executed from the Storage Manager Enterprise
Management window (select Tools

Execute Script to open this window):
//Clear Drive Channel Statistics
clear allDriveChannels stats;
//
//Reset Storage Subsystem SOC Baseline
reset storageSubsystem SOCBaseline;
//
//Reset RLS baseline
reset storageSubsystem RLSBaseline;
The IBM Support representative handling your case might request that this script be run and
that a new Collect all Support Data file be captured the next day.
Multiple drive failures
We highly recommend logging an IBM Service call whenever multiple drives fail
simultaneously. Sometimes, the root cause is clearly understood and we can be reasonably
confident that there are no underlying hardware defects. For example, if there was an
unexpected loss of power to an expansion enclosure, then this could result in multiple disks
remaining in a failed state. Those arrays that only lost RAID redundancy will remain online,
but in a degraded state, and reconstruction or copyback will start automatically when power is
restored to the enclosure. This can be observed in the Storage Manager Subsystem
Management window physical view. The missing disks first re-appear in a replaced state. The
associated logical drives return to an optimal state when reconstruction is complete without
any intervention.
Any arrays and logical drives where the outage resulted in a failed array remain in a failed
state after power is restored to the enclosure. This applies if two or more disks in the same
RAID 5 array reside in the missing enclosure. Before taking any recovery action, it is
important to understand the order in which the disks failed. Ideally, the failed disks should be
revived in the opposite order in which they failed. With a power failure affecting a single
enclosure, we can sometimes assume that all disks failed simultaneously. However, if a disk
was in a failed state prior to the power outage, then it could contain stale data and therefore
needs to be excluded from the array during recovery. Failing to do so might result in data
corruption. If there is any doubt, then the IBM Support representative can determine the order
in which the disks failed by reviewing the MEL and shell data.
Recovery actions: Multiple disks failures
To recover after multiple disks fail, perform these steps:
1. Determine the order in which the disks failed.
2. Unassign any standby hotspare drives.
3. Revive each disk starting with the drive that failed last until the associated logical drives
change from a Failed to Degraded state. With a RAID 5 array, this means that one disk still
remains in a failed state.
The Revive option is available through the Storage Manager Subsystem Management
window by highlighting the drive and then selecting Advanced

Recovery

Revive
Drive.
In this case, reviving a failed disk results in it being returned to an Optimal state.
4. Reboot the host(s) or rescan for the previously missing LUNs.