It seems that this was a hard question. Only received a reply from :
alan_at_nabeth.cxo.dec.com
Since his point of view (which i share) it is hard to map stripeset bad
blocks to member bad blocks. It looks like there is no way to accomplish
it.
So, apparently, some ways to find the wrong member are :
1. If the failing stripe is part of an LSM mirrored volume(s), then
disable the plex(es) involved (which at least let us working) , then,
from HSZ40 console delete the unit, the stripset, and add new individual
units to perform stressing read-write tests and, if you're lucky, find
out the wrong disk. Then, change it, create again the stripeset, add the
unit and enable again the previously disabled plex(es).
2. If the failing disk is not part of an LSM mirrored volume, follow the
same steps as in 1, but your data won't be available for the time it
takes to complete all the tasks. Additionally, you'll have to recover
all the data from backups
3. If the failing disk is not part of an LSM mirrored volume, and you
can't stop your system, you're in serious troubles. You may wait until
the failing member definitely goes offline or crashes, replace the
failing member and recover data from backups.
Regards
%%%%%%%%%%%%%%%%%%%%%%%%%
F. Javier Aida Nakamo
Software Engineer - GMD S.A.
jaida_at_gmd.com.pe
Tlf: +51 - 1 - 241-2626 Ext. 254
FAX: + 51 - 1 - 446-9667
Lima - PERU
%%%%%%%%%%%%%%%%%%%%%%%%%
Received on Wed Aug 21 1996 - 20:27:22 NZST