Veeam says some Software Appliance virtual machines on AMD EPYC hosts intermittently fail to boot, including after a kernel update. Operators need a recovery plan for the reboot, while still delivering security fixes to their backup infrastructure.

The vendor’s KB4960, published November 4, identifies Veeam Software Appliance 13.1 and later. Sergiu Gatlan’s November 6 BleepingComputer report brings together the boot warning and a separate Linux security update.

Evidence status: vendor-reported availability failure. The boot issue is a failed startup health check, not evidence of an attack or corrupted backup data. Veeam says normal Backup & Replication operation is unaffected once the appliance boots successfully.

A cryptographic check stops the appliance before startup

The Linux-based appliance runs Backup & Replication without a separate manual software installation. During early boot, its kernel performs a jitterentropy health test, checking a source of randomness, in FIPS mode, used for cryptographic compliance.

Veeam says the test intermittently fails on certain AMD EPYC processors, including 7262, 7302 and 7F72. Those are examples, not a complete affected-processor list. The resulting kernel panic stops startup and leaves the appliance unavailable. Depending on the hypervisor, the message may appear only in the serial console or event log.

As of the reviewed advisory, no fix is available for the kernel version used by the appliance. This refers to the boot failure; a security repair discussed below is a separate change.

Plan for a power cycle, not another guest restart

Veeam’s preferred workaround is to redeploy the appliance on a host with a different processor. Where that is impractical, its documented fallback is specific:

  1. Disable the appliance’s automatic updates.
  2. Install updates manually and monitor the virtual machine as it reboots.
  3. If this kernel panic occurs, power the virtual machine off and on again. Restarting the guest operating system does not recover it.

This is a move to supervised updates, not permission to leave the appliance unpatched. Arrange authorized access to the hypervisor console before the maintenance window. Confirm the appliance and expected backup operations return afterward.

The advisory does not recommend disabling the cryptographic health test or FIPS mode. Its documented recovery path avoids inventing a weaker startup configuration.

The Linux security update still needs to land

Veeam’s separate November 6 impact statement, KB4966, covers CVE-2026-98432, a Linux networking memory-lifetime flaw. Our existing platform-patch coverage explains the controlled exploit demonstration and why disabling multipath TCP is insufficient.

For Software Appliance and Infrastructure Appliance 13.1, Veeam says its repository has offered the repaired kernel 5.14.0-570.60.1+27.2.el9_6_ciq since November 2. Use Veeam Updater and check update history. A local repository mirror must synchronize before it can deliver the update. Complete any required restart and verify the intended kernel is running.

Other components have different delivery paths. Windows-based Backup & Replication deploys Linux helper appliances too; KB4966 specifies a private fix for 13.1.2.57, with the 12.3 fix still pending. Recovery media created before the relevant update must be replaced. Match the component to the vendor’s table rather than assuming the management server’s operating system determines every dependency.

Caveat: Veeam supplies both advisories and sells the affected software; neither establishes active exploitation of these backup installations or that the security update fixes the EPYC boot failure.