We currently have two servers with SATA disks. The box load on the box every so often jumps. We see a direct correlation between the IO wait and the system usage going up. When ever the IO is high the system CPU usage jumps as well. The server has hardware raid with the drive showing as /dev/sda. We re running CentOS. We want to install two SSD's in a raid 1 as well, boot from a USB key and then use dd to copy from the SATA disks over to the SSD's. Once we copy everything over we will remove the original array and leave just the second. Will this work? How does Linux assign which disk is /dev/sda or /dev/sdb?
As you have the
LVM partition I would recommend you the following steps:
First switch the
grub configuration to use
UUIDs for /boot volume the system uses.
Next, attach the
SSD device, boot the system from the
RAID and partition the
SSD to have two partitions: for a new /boot volume and a new
Then add that new
PV to existing
VG and migrate all
LVs to an
PV (check with
man lvm for all necessary commands).
After migration is done, remove the
PV from a
Then copy with tar or rsync the content of /boot file system to the new location (the /boot partition on
SE contexts, adjust the kernel settings in the grub configuration and install grub on