I have a system running CentOS 5.5 with 26 drives from Western Digital.
-
2 drives, sda and sdb, are hooked up to an onboard Nivdia MCP55 SATA controller, providing md0 for /boot and md1 for /
-
12 drives, sdo, are hooked up to an ARC-1231 configured as RAID6
-
12 drives, sdc through sdn, are hooked up to an ARC-1231 configured as JBOD
My plan was to do a comparision betweeen hardware and software RAID6. I initialized the 12 disks as a single partition marked as linux raid autodetect. That worked fine.
:::bash
for d in /dev/sd{c..n}; do
sfdisk -uS $d << EOF
32,2930276320,fd
EOF
done
However, whenever I tried to create the RAID device it failed.
# mdadm --create /dev/md2 --verbose --raid-devices 11 --spare-devices 1 --chunk 128 --level 6 /dev/sd{c..n}1
mdadm: layout defaults to left-symmetric
mdadm: Cannot open /dev/sdc1: Device or resource busy
mdadm: Cannot open /dev/sdd1: Device or resource busy
mdadm: Cannot open /dev/sde1: Device or resource busy
mdadm: Cannot open /dev/sdf1: Device or resource busy
mdadm: Cannot open /dev/sdg1: Device or resource busy
mdadm: Cannot open /dev/sdh1: Device or resource busy
mdadm: Cannot open /dev/sdi1: Device or resource busy
mdadm: Cannot open /dev/sdj1: Device or resource busy
mdadm: Cannot open /dev/sdk1: Device or resource busy
mdadm: Cannot open /dev/sdl1: Device or resource busy
mdadm: Cannot open /dev/sdm1: Device or resource busy
mdadm: Cannot open /dev/sdn1: Device or resource busy
mdadm: create aborted
I wasn’t able to find any processes that had the devices open, and google didn’t provide any advice that seemed relevant. Eventually I thought to check the device mapper with dmsetup ls and saw that the problem was multipathing. After flushing the unused paths with multipath -F, blacklisting the drives in /etc/multipath.conf with the stanza below, and restarting multipathd it worked fine.
device {
vendor "WDC"
}