jhetrick62
April 13th, 2010, 04:44 PM
I used 10.04 amd_x64 server install, beta 2, to build a server. I defined 4 raid-1 arrays on the primary drives so that they would be redundant in hot-swap bays. Those drives were /dev/sda and /dev/sdc and each were 500gb drives with identical partitions for the 4 raid 1 devices including a mirrored swap drive. The machine ran fine and even survived rebooting several times. (I upgrade parted before building this as I found the bug report about not being able to create a drive above 2tb)
After loading the system, I then built a large storage raid-0 array which included (4) 1tb drives. These were /dev/sdb /dev/sdd /dev/sde and /dev/sdf. After assembly of the array, it was reflected properly in /proc/mdstat. I did not format this yet, but rebooted.
Upon reboot, the raid device md4 was now listed as md_d4 and it was inactive. To make matters worse, it listed the (4) active raid-1 arrays as all consisting of devices /dev/sdcx and /dev/sdex ! Why not /dev/sda and /dev/sdc as originally built. Mind you that other than loosing my large array, the machine is runnnig fine and it runs on the (4) raid-1 arrays as they are for /, /home, swp and one other that I use for vm appliances called /virtbox.
I then ran the --stop command on the array and ran the --remove command on the array and then rebooted. It now still lists an inactive array of md_d4 BUT the original (4) raid devices are back to on sda and sdc, go figure?
I assume that the changing drive assignements are messing with the raid build of the md4 array with the (4) 1tb drives.
Any suggestions?
After loading the system, I then built a large storage raid-0 array which included (4) 1tb drives. These were /dev/sdb /dev/sdd /dev/sde and /dev/sdf. After assembly of the array, it was reflected properly in /proc/mdstat. I did not format this yet, but rebooted.
Upon reboot, the raid device md4 was now listed as md_d4 and it was inactive. To make matters worse, it listed the (4) active raid-1 arrays as all consisting of devices /dev/sdcx and /dev/sdex ! Why not /dev/sda and /dev/sdc as originally built. Mind you that other than loosing my large array, the machine is runnnig fine and it runs on the (4) raid-1 arrays as they are for /, /home, swp and one other that I use for vm appliances called /virtbox.
I then ran the --stop command on the array and ran the --remove command on the array and then rebooted. It now still lists an inactive array of md_d4 BUT the original (4) raid devices are back to on sda and sdc, go figure?
I assume that the changing drive assignements are messing with the raid build of the md4 array with the (4) 1tb drives.
Any suggestions?