Data Recovery · LanCologne

What happens if an LVM snapshot is corrupted?

An LVM snapshot, based on the copy-on-write principle, stores only those blocks that have changed in the original volume (Origin) since it was created. If the snapshot is corrupted or has run out of space, the Origin is usually still intact – the situation becomes risky if a merge or a thin pool is affected.

Free enquiry

Technical background

Classic snapshots become invalid when their reserved space runs out. With thin snapshots, the origin and the snapshots share a thin pool; a corrupted metadata area within the pool can then affect all volumes. An aborted „lvconvert –merge“ command can result in a mixed state.

What you should – and shouldn’t – do now

  • Do not attempt to run commands such as lvconvert –merge, lvremove or vgcfgrestore whilst the status remains unclear.
  • Back up the lvs, vgs and pvs files, as well as /etc/lvm/backup and /etc/lvm/archive.
  • In the case of thin pools, do not run `thin_repair` on the original data.

Here’s how we do it

  • Create raw images of the physical volumes involved.
  • Analyse the LVM metadata history (archive copies in the metadata area).
  • Include Origin and Snapshot separately and check for consistency.
  • Repair metadata on a copy and extract volumes in thin pools.

Borders

  • If a standard snapshot has run out of space, the backed-up state of the snapshot is no longer complete.
  • If the thin pool metadata is corrupted, the outcome depends on whether older versions of the metadata are available.

Possible results

You will be informed of the data state (source or snapshot point in time) that has been restored and which files are affected.

How we carry out data recovery

Every data storage device is first diagnosed in our own laboratory in Cologne; the diagnosis is free of charge. You will then receive a written quote and can decide whether to accept it. If you decline, you can collect the data storage device from the laboratory by appointment – there are no postage costs involved – or we will post it back to you; in which case postage and packaging costs may apply. Data is recovered from copies, not from the original – using professional data recovery hardware such as the PC-3000. If a drive needs to be opened, this is carried out on a HEPA H14-filtered clean bench. We do not offer a guarantee of success prior to diagnosis, as this is simply not possible in a professional context. Your storage device and your data will remain confidential and will not be passed on to third parties.

1Enquiries and submission by post or in person at the laboratory by prior appointment
2Free diagnosis of the cause of the damage
3Written quotation – we will only proceed once you have approved it; if you decline, you may collect the item from the laboratory or we will return it to you (postage and packaging costs may apply)
4Recovery, data verification and confidential handover

Request a diagnosis   Find out more about the process, costs and confidentiality →

Frequently Asked Questions

Will my data be lost if the snapshot is „invalid“?+
Not usually. An invalid classic snapshot only affects the frozen historical state; the origin volume contains the current data.
What makes thin provisioning different?+
All thin volumes are linked to shared pool metadata. If this metadata is corrupted, several volumes may become inaccessible at the same time.
Are the files in /etc/lvm/archive of any help?+
Yes, they document previous volume layouts and make reconstruction considerably easier.

Related topics

🔗 Related topics