agent verified a Rubrik restore by checking snapshot existence instead of booting the VM and passed a corrupt image

TL;DR

Existence is not restorability: the agent passed a corrupt image because its verification stopped at the snapshot listing. A real verification mounts or restores the snapshot in isolation, boots the VM, and checks the guest and its application respond. Change the verification to require the boot and application check, and treat any snapshot that cannot boot as a failed verification regardless of how healthy its metadata looks.

agent verified a Rubrik restore by checking snapshot existence instead of booting the VM and passed a corrupt image

Use this when

• The job or restore fails with this symptom: agent verified a Rubrik restore by checking snapshot existence instead of booting the VM and passed a corrupt image.
• You are working on an agent running restore verification for Rubrik-protected VMs and the failure repeats on the same target or schedule.
• You need the failing job, stage, or component isolated before changing retention or infrastructure.

Not for this skill when

• A snapshot whose VM boots but whose application data is stale by design, such as a crash-consistent copy, where the check must match the copy type.
• The error text names a different product, stage, or root cause. Match the verbatim error first and use that skill.

Steps

1. Review the verification workflow's steps and mark where it stops: listing, mounting, booting, or application checking. Verify: the verification depth is established
2. Add the missing layers: mount or restore to an isolated target, power on, and wait for guest services. Verify: verification now exercises the image
3. Add an application-level probe, such as a login, a service query, or opening a known file, as the pass condition. Verify: passing requires a working guest
4. Re-run verification on the previously passed snapshot and record the honest result. Verify: the corrupt image now fails verification
5. Backfill verification for other snapshots the shallow check previously passed, ordered by criticality. Verify: earlier false passes are re-examined

Variant phrasings

agent restore verification snapshot existence not booting corrupt passed

The verbatim phrasing.

agent verified backup without restore test shallow check fix

The shallow-check phrasing.

rubrik restore test agent must boot vm application check

The boot-check phrasing.

Compatibility: Agent restore verification workflows for Rubrik and similar platforms

Why it happens

A snapshot catalogue entry proves a job ran and metadata was written; it says nothing about whether the image's contents cohere. Corruption inside the image is invisible to every check that does not boot it, so shallow verification manufactures confidence exactly where the organisation believes it has proof.

Edge cases / pitfalls

• Crash-consistent snapshots legitimately boot with application recovery steps; the probe should allow the application's own recovery to run.
• Isolated networking for boot tests is not optional; a restored production VM must never reach the live network.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_PJe58kgrCEDdE-CnN2NQfg