Active logical volumes using Azure snapshots as backing storage show I/O errors for no apparent reason

Issue

• Currently-active logical volumes (LVs) hosted on Azure snapshot devices report I/O errors, despite no sign of errors on the underlying storage disks:

[  495.545481] Buffer I/O error on dev dm-XX, logical block 0, async page read

• LVM commands report no issues with the logical volumes, and show them as active.

• Affected LVs do not show any typical signs of LVM-layer issues (no issues with thin pool metadata, no missing PVs, no errors reported by LVM commands, etc).

• Affected LVs cannot be mounted, and no longer show any kind of filesystem signature on them, despite previously showing no issues:

[root@example1 ~]# mount /dev/mapper/VG01-LV01 /mnt
mount: /mnt: can't read superblock on /dev/mapper/VG01-LV01.

[root@example1 ~]# blkid /dev/mapper/VG01-LV01
[root@example1 ~]#