Veritas VXFS corruption seen on Linux VM with RDM LUNs
book
Article ID: 448216
calendar_today
Updated On:
Products
VMware vSphere ESXi
Issue/Introduction
Symptoms:
A Linux Virtual Machine (VM) utilizing Raw Device Mapping (RDM) LUNs may experience Veritas File System (VXFS) corruption. This issue is typically identified by specific kernel messages regarding inode allocation failures, leading to the filesystem being flagged for a mandatory consistency check.
The Guest OS filesystem becomes read-only or inaccessible.
Application failures (e.g., Veritas Cluster Server or databases) seen from Guest OS -
Jul 12 03:33:41 gbl20054217 kernel: vxfs: msgcnt 12 mesg 010: V-2-10: vx_ialloc - /dev/vx/dsk/<Disk Group>/<Logical Volume> file system inode 443 not free Jul 12 03:33:41 gbl20054217 kernel: vxfs: msgcnt 13 mesg 016: V-2-16: vx_ilisterr: vx_ialloc - /dev/vx/dsk/<Disk Group>/<Logical Volume>file system error reading inode 443 Jul 12 03:33:41 gbl20054217 kernel: vxfs: msgcnt 14 mesg 096: V-2-96: vx_setfsflags - /dev/vx/dsk/<Disk Group>/<Logical Volume>file system fullfsck flag set - vx_ilisterr_1. Please run fsck to clean the filesystem. To collect metasave add -o metasave option while running fsck
Environment
VMware vSphere ESXi
Cause
The filesystem corruption originates within the Guest OS layer. When using RDM LUNs, the ESXi hypervisor bypasses the VMFS layer and is only responsible for LUN liveness and pathing. The integrity of the filesystem metadata is managed entirely by the Guest OS and the Veritas VXFS. The corruption is frequently attributed to software-level inconsistencies within the Guest OS filesystem stack.
Resolution
Engage Guest OS to check the filesystem corruption.
Additional Information
For RHEL-specific metadata I/O errors on RDMs, see KB 402858.