HP SmartArray controller disk replacement steps on HP i2 blades

This is for the BL860c i2 blades (and other i2 models):

Replace a failed LVM mirror disk that is attached to a SmartArray SAS controller.


Notes:  For this example /dev/disk/disk2 is the failed disk and /dev/disk/disk4 is the new device that was created because the SAS controller identified a new disk.

Notes:  For this example the SAS controller device is /dev/ciss0

Notes:  The failed disk is in volume group vg00


Identify the disk by serial number.  Give the Bay number to the HP CE so they know which disk to replace:

sautil /dev/ciss0 -s

scsimgr get_info -D /dev/rdisk/disk2


Deactivate the PV:

pvchange -a n /dev/disk/disk2


Replace the failed disk.


Activate the WWID of the new disk.  The Error can be ignored.  A new disk device should be created:

scsimgr -f replace_wwid -D /dev/rdisk/disk2 dsf

scsimgr: ERROR: Cannot convert hardware path to an IO tree node


Find the new disk device that was created:

ioscan -fNnC disk


Create a description file to create the EFI partitions:

# cat >> /tmp/idf << EOF



HPUX 100%



Use idisk to setup the disk partitioning on the new disk device:

idisk -wf /tmp/idf  /dev/rdisk/disk4

insf -e –C disk


Assign the old instance number to the replacement disk:

io_redirect_dsf -d /dev/disk/disk2 -n /dev/disk/disk4


Verify that /dev/disk/disk2 is in a claimed state:

ioscan –fnNC disk


Use mkboot to format and populate the newly created EFI partition:

mkboot -e -l /dev/disk/disk2


Change the AUTO file contents to the proper mode:

mkboot -a "boot vmunix -lq" /dev/rdisk/disk2


Restore the LVM reserved areas (PVRA/VGRA):

vgcfgrestore –n vg00 /dev/rdisk/disk2_p2


Reattach the new disk:

pvchange -a y /dev/disk/disk2_p2


Sync volume group:

vgsync vg00


Use lvlnboot to ensure that the LVM logical volumes are prepared:

lvlnboot -R

lvlnboot -v

SAS controller on HPUX itanium disk replace

The SAS controller in the first generation BL860C (and other models) has some special commands in order to replace a failed disk.  I would recommend using the built-in RAID-1 capabilities of this controller - that way you no longer need to worry about LVM mirroring.  But if you need to use LVM mirroring then, add the couple steps below to your standard LVM mirror root/boot disk procedure:

#get the SAS controller dsf:
saslist get_info ctrl sasd

#get info on the disks:
sasmgr get_info -D /dev/sasd0 -q raid

#if the disk is working enough to turn on the attention LED:
sasmgr set_attr -D /dev/sasd0 -q lun=/dev/dsk/c0t1d0 -q locate_led=on

AFTER the disk is replaced:
#since each disk will have a unique WWID, the server will assign a new HW path and persistent dsf
#to change them first change the hardware path (legacy style)
sasmgr replace_tgt -D /dev/sasd0 -q old_tgt_hwpath=0/2/1/ -q new_tgt_hwpath=0/2/1/

#then change the persistent dsf
io_redirect_dsf -d /dev/disk/disk3 -n /dev/disk/disk8