How to Migrate iSCSI Multipath Storage Between Linux Servers Without Downtime

Scenario

A friend has Server A that accesses a storage array via iSCSI multipath. Eight 58 TB disks are combined with multipath, then merged into a single LVM volume group and formatted with XFS, mounted at /var/www. The goal is to detach this storage from Server A and attach it to Server B.

Step‑by‑Step Migration

Stop all applications on Server A.

Kill any processes using the mount point: fuser -ck /var/www Unmount the filesystem: umount /var/www Disconnect all iSCSI sessions: iscsiadm -m node -U all Delete all node records: iscsiadm -m node --op delete On Server B, install multipath-tools and open‑iscsi, then copy the multipath.conf from Server A to /etc/.

Connect the iSCSI target to Server B and update the IQN of the initiator in the storage mapping.

Run discovery and login commands (replace <storage_ip> with the actual target IP):

iscsiadm -m discovery -t st -p <storage_ip>
iscsiadm -m node -p <storage_ip> -l

Enable and start the iSCSI service:

systemctl enable iscsi.service
systemctl start iscsi.service

Set node startup to automatic:

iscsiadm -m node -o update -n node.startup -v automatic

Restart multipathd so the new LUNs are recognized.

Rescan LVM: lvscan The logical volumes should appear automatically; verify with vgs, pvs, and lvs.

Edit /etc/fstab to add the new mount point, then apply with:

mount -a

Observed Issue

After the migration, running mount -a on Server B reports a superblock error, yet a manual mount works: mount /dev/mapper/datavg-datalv /mnt The filesystem is readable and writable, so the error is puzzling.

Troubleshooting

Running xfs_repair -n /dev/mapper/datavg-datalv shows no problems. Checking the kernel log with dmesg | tail reveals the message: unknown mount option [defautls] The /etc/fstab entry contains a typo: the mount option defaults was misspelled as defautls. Correcting the spelling eliminates the superblock error and allows mount -a to succeed.

Conclusion

The migration itself is straightforward, but a single character mistake in a configuration file caused a seemingly serious failure. Always back up configuration files and double‑check entries such as /etc/fstab before applying changes to avoid costly downtime.