Hi,
I think I understand this but I just want to run it past you guys
as the consequences of my getting it wrong in this case are
too horrible to imagine:
If in an HSZ disk array (RAIDset or MIRRORset), a disk
goes south, it joins the FAILEDset and a disk from the
SPAREset takes over. So far so good. But when we take
out the bad disk (after having done DELETE FAILEDset
diskname) and pop in a new one, what do we do now to
make this new disk take over the role that the old disk had
before it died? It is (allegedly) important that we have the
disk physically in the right slot so we can maximize performance
by striping the RAIDset or MIRRORset properly, plus
then we won't forget which disk is which.
If think we delete from the MIRROR/RAIDset the disk
from the spareset that took over and then do:
set RAIDsetname policy=nopolicy (to stop it from just subbing another
disk from the spareset when we do the remove)
set RAIDsetname remove=diskfromspareset
set RAIDsetname replace=newdiskreplaingoldfaileddisk
set RAIDsetname policy=best_performance (or whatever the policy was
before)
Am I right?
Does it make a difference if the new disk is not really a new
disk but the same old disk from the FAILEDset that the
engineer assures us is fine and we should try again ? Does
the HSZ keep a record of the disk serial number or does
it it write something to the disk to the effect that this is
a failed disk and must at least be reinitialized?
Thanks! Again I will summarize.
John
Received on Sat Aug 21 1999 - 01:31:11 NZST