Explanation
The cluster maintains a virtual medium error table
for each MDisk. This table is a list of logical block addresses on
the managed disk that contain data that is not valid and cannot be
read. The virtual medium error table has a fixed length. This error
event indicates that the system has attempted to add an entry to the
table, but the attempt has failed because the table is already full.
There
are two circumstances that will cause an entry to be added to the
virtual medium error table:
- FlashCopy, data migration and mirrored volume synchronization
operations copy data from one managed disk extent to another. If the
source extent contains either a virtual medium error or the RAID controller
reports a real medium error, the system creates a matching virtual
medium error on the target extent.
- The mirrored volume validate and repair process has the option
to create virtual medium errors on sectors that do not match on all
volume copies. Normally zero, or very few, differences are expected;
however, if the copies have been marked as synchronized inappropriately,
then a large number of virtual medium errors could be created.
User Response
Ensure that all higher priority errors are fixed
before you attempt to resolve this error.
Determine whether
the excessive number of virtual medium errors occurred because of
a mirrored disk validate and repair operation that created errors
for differences, or whether the errors were created because of a copy
operation. Follow the corresponding option shown below.
- If the virtual medium errors occurred because of a mirrored disk
validate and repair operation that created medium errors for differences,
then also ensure that the volume copies had been fully synchronized
prior to starting the operation. If the copies had been synchronized,
there should be only a few virtual medium errors created by the validate
and repair operation. In this case, it might be possible to rewrite
only the data that was not consistent on the copies using the local
data recovery process. If the copies had not been synchronized, it
is likely that there are now a large number of medium errors on all
of the volume copies. Even if the virtual medium errors are expected
to be only for blocks that have never been written, it is important
to clear the virtual medium errors to avoid inhibition of other operations.
To recover the data for all of these virtual medium errors it is likely
that the volume will have to be recovered from a backup using a process
that rewrites all sectors of the volume.
- If the virtual medium errors have been created by a copy operation,
it is best practice to correct any medium errors on the source volume
and to not propagate the medium errors to copies of the volume. Fixing
higher priority errors in the event log would have corrected the medium
error on the source volume. Once the medium errors have been fixed,
you must run the copy operation again to clear the virtual medium
errors from the target volume. It might be necessary to repeat a sequence
of copy operations if copies have been made of already copied medium
errors.
An alternative that does not address the root cause is to
delete volumes on the target managed disk that have the virtual medium
errors. This volume deletion reduces the number of virtual medium
error entries in the MDisk table. Migrating the volume to a different
managed disk will also delete entries in the MDisk table, but will
create more entries on the MDisk table of the MDisk to which the volume
is migrated.
Possible Cause-FRUs or other: