On Thu, Mar 13, 2025 at 02:08:21PM +0100, Yassine Chaouche wrote:
Le 3/12/25 à 23:11, Michael Stone a écrit :
Two of the drives are dead, you're not going to see anything from them
So this means I can't rely on smartctl to list physical disks,
I guess I don't understand how you expect smartctl to query a dead disk.
It's dead, that means it's not going to respond.
A tool that queries the structure of the raid can tell you what drives
the raid expects to see, regardless of whether they are available or
not. A tool that queries drives can't tell you that there's supposed to
be a drive that's not there.
for servers I didn't install,
I need to know their RAID setup in advance,
or at least the total number of disks,
then compare with the output of smartctl or cciss_vol_status.
Yeah. Looking again, I misread the cciss_vol_status output (which is
truly horrible). I don't see any way to determine which drives are in
which volumes from that output, and the formatting/indentation suggests
all of the drives are in the last volume listed. I guess this tool is
intended to give a quick overview, and not really manage an array. You
can look for the (non-free) ssacli tool from HPE which can actually be
used to create/query/etc an array in much more detail. I'm not aware of
a free tool for this.
If you're just looking to figure out whether there's a dead drive the
ccs_vols_status tool will do that (any volume whose status is not "OK").
smartctl won't help you, but I'm not sure why you're trying to use
smartctl at all here--in general, stuff in a hardware raid aray needs to
be managed at the array level, not at the disk level.
