Hi all,
Martin Petder corrected my (dangerous) summary - there should be:
init stripeset NODESTROY (valid for mirror and RAID 0+1)
=========
so as not to lose data. I have checked the manuals for 8.5 and 8.6 and really - the default value is DESTROY. But in spite of using the default value of INIT (i.e. DESTROY), the data have not been lost. Nevertheless - don't follow our method and use "init stripeset NODESTROY"
Martin
-----Original Message-----
From: Dušek Martin
Sent: Saturday, August 10, 2002 4:35 AM
To: tru64-unix-managers_at_ornl.gov
Subject: SUMMARY: HSG80 unit inoperative
Hi all,
I didn't receive any response so far, maybe because of our Exchange Server failure.
Fortunately, our Czech Digital support engineer (thank you, Pavel) repaired the failed unit this way:
delete unit
delete stripeset
delete mirror
add mirror
add stripeset
init stripeset (!!!) - there should be NODESTROY here
add unit
and all the original data in the stripeset have been accessible again.
Any other procedure including the same but without "init stripeset" didn't help because of corrupted metadata.
Martin
-----Original Message-----
From: Dušek Martin
Sent: Saturday, August 10, 2002 1:45 AM
To: tru64-unix-managers_at_ornl.gov
Subject: HSG80 unit inoperative
Hi, folks
I've got very urgent issue.
During controllers replacement one unit went to "inoperative" state, but with no lost data , therefore is no possible make "clear_errors ...." for starting unit in operating mode.
Unit has stripe from two mirrors ( m6,m7 ) which one of them (m6)is in inoperative state.
I tried :
1) remove one disk from inoper.mirr.(m6), removed - status doesn't changed - no effect
2) add disk back to m6 - not possible -> mirr. m6 in reduce mode
3) restart controllers - no effect
m6
1. disk - "INOPERATIVE"
(now remowed - reduced mirror)
2. disk - "no information available"
!! URGENT !!! - IN MINUTES
I need to save more than 130GB.
thanks for any ideas.
md.
Received on Mon Aug 12 2002 - 15:13:02 NZST