vCenter Appliance fsck failed

Recovering a vCenter Server Appliance Stuck in a fsck Boot Loop

If a vCenter Server Appliance (VCSA) gets stuck in a repetitive reboot loop with filesystem check (fsck) errors during boot, this is almost always disk corruption on one of the appliance’s internal partitions — commonly triggered by an unclean shutdown, such as a host losing power or the VM being forcibly reset mid-write. To recover it, boot the appliance and interrupt it at the GRUB menu (or use “Connect to Console” from vSphere/ESXi directly during the failed boot) to drop into a shell, or boot into the appliance’s rescue/recovery environment if one is offered. From a root shell, run a filesystem check against the affected partition:

e2fsck -y /dev/sda3

The -y flag automatically answers yes to every repair prompt, which is what you want for an unattended recovery pass. The exact partition (/dev/sda3 in most cases, but check which one the boot error actually names) can vary depending on your VCSA version and deployment size, so always check the specific error message rather than assuming it’s always sda3. Once e2fsck finishes reporting it has fixed all outstanding errors, reboot the appliance normally. It should come up cleanly. If the same corruption recurs shortly afterwards, that’s usually a sign of an underlying storage problem rather than a one-off VCSA fault, and it’s worth checking your host and datastore health before writing it off as resolved.

Still Current in 2026, Plus a Licensing Note

This procedure remains valid and is still Broadcom’s own documented method for current VCSA versions, including 9.x — the appliance still runs a Photon OS ext4 root filesystem with the same recovery boot workflow, so the steps above haven’t changed. Worth knowing if you’re troubleshooting this today: since the Broadcom acquisition, VMware has moved to subscription-only licensing and discontinued perpetual licenses and their associated support contracts, so confirm your vCenter is still under an active subscription with support entitlement before opening a case. If it isn’t, Broadcom’s own knowledge base articles and the community forums remain the practical self-service fallback for this kind of recovery.

Resources

• VMWare