418 IBM Midrange System Storage Hardware Guide
It is only when data on the sector becomes unreadable that the DS4000 or DS5000 controller
must intercept by filling in the gaps using RAID redundancy data from the remaining disks in
the array. If an error threshold is exceeded, then the disk is marked as failed and spun down
awaiting replacement. At this point, if the array has RAID redundancy, there needs to be a
reconstruction of data either to a standby hotspare drive, if available, or onto a replacement
disk. With RAID 1 arrays, this involves a full copy of every sector for all logical drives from just
the partner disk in the mirrored pair. With RAID 3 or 5 arrays, this involves a full copy of every
sector for all logical drives from data and parity information about every remaining disk in the
array.
Although extremely rare, it is nonetheless possible that an unreadable sector is detected on
one of the source disks during this reconstruction. With RAID 3 and 5 arrays, the risk
increases proportionally with the number of disks in the array. The type and capacity of the
disks can also be a factor. RAID 6 arrays have two parity disks, so the potential exposure to
this problem is statistically negligible.
When this occurs, there is no way to recover the data from the affected sector(s). The
DS4000 or DS5000 controllers then remap the bad block without taking a prior copy of the
data, while appending a log entry to the unreadable sector list. This allows the reconstruction
process to complete, but data on the bad block is lost. Recovery Guru notifies the user of a
non-optimal condition whenever there are any entries in the unreadable sector list. This list
can be accessed from the Storage Manager Subsystem Management window by selecting
Advanced Recovery Unreadable Sectors. By reviewing the logical LBAs in the list, it
might be possible to determine whether the bad block(s) are in free space or a file. This could
assist any decision on recovery action.
The DS4000 or DS5000 unit remains in a non-optimal condition until the Clear option is
selected in the Unreadable Sectors window. However, make sure that a Collect All Support
Data file is captured before doing so.
7.8.2 Diagnosing host-side problems
A significant proportion of support calls logged for host-side problems are found to be due to
either misconfiguration or faults external to the DS4000 or DS5000. In this section, we look at
some areas to check when attempting to identify host related problems.
Some of the more common host-side problems on the DS4000 or DS5000 include:
Logical drive not on preferred path
LUN bouncing
Persistent reservations
Target reset
Checking host configuration rules
The host configuration rules are described in detail in the IBM System Storage DS Storage
Manager Version 10 Installation and Host Support Guide, GC53-1135. The latest operating
system and code version specific updates are available in the readme files that accompany
the Storage Manager download files. You can find these files at the following address:
http://www.storage.ibm.com/support
Supported configurations can be checked at the System Storage Interoperation Center at the
following address:
http://www-03.ibm.com/systems/support/storage/config/ssic
To Next Page
Loading...
Need help?
General