Perform this procedure to enable uan_disk_config to run successfully by erasing existing disk partitions. UAN disk configuration will fail if the disk on the node is already partitioned. Manually erase any existing partitions to fix the issue.
This procedure currently only addresses uan_disk_config errors due to existing disk partitions.
Refer to Basic UAN Configuration for an explanation of UAN disk configuration.
The most common cause of failure in the uan_disk_config role is the disk having been previously configured without a /scratch and /swap partition. Existing partitions prevent the parted command from dividing the disk into those two equal partitions. The solution is to log into the node and run parted manually to remove the existing partitions on that disk.
Examine the CFS log and identify the failed disk device.
Log into the affected UAN as root.
Use parted to manually remove any existing partitions.
The following example uses /dev/sdb as the disk device. Also, as partitions are removed, the remaining partitions are renumbered. Therefore, rm 1 is issued twice to remove both partitions.
uan# parted
GNU Parted 3.2
Using /dev/sda
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) select /dev/sdb
Using /dev/sdb
(parted) print
Model: ATA VK000480GWSRR (scsi)
Disk /dev/sdb: 480GB
Sector size (logical/physical): 512B/4096B
Partition Table: msdos
Disk Flags:
Number Start End Size Type File system Flags
1 1049kB 240GB 240GB primary ext4 type=83
2 240GB 480GB 240GB primary ext4 type=83
(parted) rm 1
(parted) rm 1
(parted) print
Model: ATA VK000480GWSRR (scsi)
Disk /dev/sdb: 480GB
Sector size (logical/physical): 512B/4096B
Partition Table: msdos
Disk Flags:
(parted) quit
uan01:~ #
Either reboot the affected UAN or launch a CFS session against it to rerun the uan_disk_config role.