Nimble storage devices periodically do not form multipath devices
Issue
- After booting, Nimble storage devices are expected to form a multipath device, but when the system fully boots they did not.
Expected:
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:16 0 745.2G 0 disk
|-sda1 8:17 0 1G 0 part /boot
`-sda2 8:18 0 550.8G 0 part
|-rhel-root 253:0 0 293G 0 lvm /
`-rhel-swap 253:1 0 62.5G 0 lvm [SWAP]
sdb 8:32 0 7.6T 0 disk
`-mpatha 253:3 0 7.6T 0 mpath
sdc 8:48 0 7.6T 0 disk
`-mpatha 253:3 0 7.6T 0 mpath
Actual:
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:16 0 745.2G 0 disk
|-sda1 8:17 0 1G 0 part /boot
`-sda2 8:18 0 550.8G 0 part
|-rhel-root 253:0 0 293G 0 lvm /
`-rhel-swap 253:1 0 62.5G 0 lvm [SWAP]
sdb 8:32 0 7.6T 0 disk
sdc 8:48 0 7.6T 0 disk
-
Systemd kills either
multipathd.service
, orlvm2-activation-early.service
during the boot process. Ifmultipathd.service
is the service that is killed, multipaths will never be formed.Feb 3 16:40:11 localhost systemd: Found ordering cycle on lvm2-activation-early.service/start Feb 3 16:40:11 localhost systemd: Found dependency on multipathd.service/start Feb 3 16:40:11 localhost systemd: Found dependency on nlt.service/start Feb 3 16:40:11 localhost systemd: Found dependency on lvm2-monitor.service/start Feb 3 16:40:11 localhost systemd: Found dependency on lvm2-activation.service/start Feb 3 16:40:11 localhost systemd: Found dependency on lvm2-activation-early.service/start Feb 3 16:40:11 localhost systemd: Breaking ordering cycle by deleting job multipathd.service/start Feb 3 16:40:11 localhost systemd: Job multipathd.service/start deleted to break ordering cycle starting with lvm2-activation-early.service/start
Environment
- Red Hat Enterprise Linux 7
- Device-Mapper-Multipath (DM-multipath)
- Logical Volume Manager (LVM2)
- HPE Nimble Storage Linux Toolkit (NLT)
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.