How to run fsck with LVM

Question

I am having I/O errors on /dev/sda on a RHEL 9 server.

Output of dmesg | grep sda:

[ 3.323141] sd 0:0:0:0: [sda] 62914560 512-byte logical blocks: (32.2 GB/30.0 GiB)
[ 3.323174] sd 0:0:0:0: [sda] Write Protect is off
[ 3.323177] sd 0:0:0:0: [sda] Mode Sense: 3b 00 00 00
[ 3.323218] sd 0:0:0:0: [sda] Write cache: disabled, read cache: disabled, doesn't support DPO or FUA
[ 3.327168] sda: sda1 sda2 sda3
[ 3.327468] sd 0:0:0:0: [sda] Attached SCSI disk
[ 6.338087] XFS (sda2): Mounting V5 Filesystem fb24877b-278d-412a-8f68-e7f1909484b1
[ 6.390410] XFS (sda2): Ending clean mount
[ 253.414807] sd 0:0:0:0: [sda] tag#58 task abort on host 0, 0000000030035026
[ 253.415338] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_TIME_OUT driverbyte=DRIVER_OK cmd_age=180s
[ 253.415342] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 01 ac b3 20 00 00 08 00
[ 253.415343] I/O error, dev sda, sector 28095264 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2

However, this disk uses LVM. Should the fsck command be run on the underlying disk /dev/sda, or on the LVM partitions /dev/mapper/...?

• Marcus Müller: Uh-oh: Your sda is broken: [ 253.415343] I/O error, dev sda, sector 28095264 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 2 fsck can't rescue that – you need to replace the hardware and hope you haven't lost any importnat data.

Answer 1

fsck checks file systems, so it needs to be run on volumes containing file systems; on LVM, you need to find the appropriate logical volume, which you’ll find in /dev/mapper and in the volume group directory under /dev. If you’re using encryption on top of LVM (that is to say, encrypting individual volumes), then you need to find the appropriate decrypted volume in /dev/mapper.

Perhaps the simplest way to do this is to use whatever is specified in /etc/fstab or in mount’s output.

(Incidentally, fsck is unlikely to help with the errors you’re seeing; it’s probably time to replace the storage.)

• dr_: This makes sense -- running a fsck on /dev/sda returns bad superblock errors.
• dr_: "it’s probably time to replace the storage" It's a VMware virtual disk, how do you advise to proceed in this case?
• Stephen Kitt: The virtual disk lives on physical storage somewhere, that’s where the errors are coming from, and that’s what needs to be checked and probably replaced.