Marcus Sundman wrote:
> "James C. McPherson" <[EMAIL PROTECTED]> wrote:
>> Marcus Sundman wrote:
>>> I couldn't figure out which controller got which numbers so I had
>>> to disconnect drives one by one
>> I'm interested in what you did to figure out your drive
>> locations - did you use cfgadm, fmtopo or sestopo to figure
>> it out, or something closer to the hardware?
>
> Disconnect one drive, boot, look which drive is missing, shutdown,
> reconnect it and disconnect the next drive, boot, etc.
urk :(
There's are less power-cycle-y ways, if you're ok with wandering
through the device tree.
For each device that you've got, try this::
$ for dev in `awk -F'"' '/sd/ {print $2}' /etc/path_to_inst`; do \
prtconf -v /devices/$dev|egrep -i "id1|dev.dsk.*s2" ; \
done
That'll tell you the cXtYdZ and device serial number mappings, ala
blinder:jmcp $ for dev in `awk -F'"' '/sd/ {print $2}' /etc/path_to_inst`;
do prtconf -v /devices/$dev|egrep -i "id1|dev.dsk.*s2" ; done
value='id1,[EMAIL PROTECTED]'
dev_link=/dev/dsk/c2t0d0s2
value='id1,[EMAIL PROTECTED]'
dev_link=/dev/dsk/c2t1d0s2
value='id1,[EMAIL PROTECTED]'
dev_link=/dev/dsk/c2t2d0s2
value='id1,[EMAIL PROTECTED]'
dev_link=/dev/dsk/c2t3d0s2
Note that the "id1" is the start of the devid string, which for SATA
devices like mine above (attached to a SAS hba)
Conversely, on my Dell laptop with AHCI, I find that using cfgadm is
more useful::
$ cfgadm -lav sata0
Ap_Id Receptacle Occupant Condition Information
When Type Busy Phys_Id
sata0/0::dsk/c1t0d0 connected configured ok Mod:
SAMSUNG HM320JI FRev: 2SS00_01 SN: S19FJ10PC45360
unavailable disk n /devices/[EMAIL PROTECTED],0/pci1028,[EMAIL
PROTECTED],2:0
sata0/1 empty unconfigured ok
unavailable sata-port n /devices/[EMAIL PROTECTED],0/pci1028,[EMAIL
PROTECTED],2:1
Having the physical serial number reported makes it much easier
to match the physical drive with the logical - as long as you've
got the lid off :-)
>> What sort of hardware, btw?
>
> A motherboard with 6 SATA ports (I think 4 is by the nvidia chipset and
> 2 by some other chip) and a sil3132-based PCI-E card with 2 SATA ports.
> The disks are 1 TB WD GP ones.
>> Oh, and if you gave the zpool create command the cXtYdZ
>> name, then the labels will have been over-written with GPT
>> (aka EFI) labels instead.
>
> Yes, I used the cXtYdZ syntax with zpool create, but the drives already
> had EFI labels. Did zpool re-label them anyway?
I believe so.
James C. McPherson
--
Senior Kernel Software Engineer, Solaris
Sun Microsystems
http://blogs.sun.com/jmcp http://www.jmcp.homeunix.com/blog
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
