KVM guest backup issue in Cloudmin

Good afternoon,

We have a Cloudmin Pro subscription.

We have a series of automated backups that run to a NAS every night configured in Cloudmin. They run over SSH.

These backups work 100%, with the exception of a new VM we've deployed. This VM is a clone of a template VM we've been using for multiple VMs now, and those seem to backup without issue.

The error we see in the backup log is:

Failed : Snapshot of /dev/vg_vmdata/emq_virtual_img failed : Using default stripesize 64.00 KiB

I have also attached a screencap for reference. We're not clear on how to assess & fix this, can you please offer some advice and/or feedback as to what the root cause of the issue may be and how to address it?

Thank you kindly in advance,

Chad

Status: 
Closed (fixed)

Comments

Was this problem VM created using Cloudmin's cloning feature, or was the original VM converted to an image and the new VM then created from that image?

chadwick89's picture
Submitted by chadwick89 on Mon, 04/30/2018 - 12:01

Hi Jamie,

So upon further testing, here's the info I have in response:

  1. This VM (VM-A) was a clone of another production VM we have (VM-B).
  2. The VM (VM-B) it was cloned from was also a clone of an older template VM (VM-C).
  3. The original template VM (VM-C) no longer exists (replaced with a new more solid one).
  4. Backups of the production VM (VM-B) it was cloned from work fine.

We just created a new system backup and selected the radio button for:

Shut down systems? Yes (ensures filesystem consistency)

This backup worked fine, so it seems isolated to the LVM snapshot.

Ok, that's expected (sort-of). A VM created by cloning another is done using an LVM snapshot, but this means that it cannot itself be snapshotted for an online backup.

chadwick89's picture
Submitted by chadwick89 on Tue, 05/01/2018 - 17:48

Thanks for the feedback. We have numerous other systems that are cloned from the same template, and their backups seem to work fine. Given that this issue could have impacts on future systems deployed, is there any way to troubleshoot this? Is deploying VMs using the cloning process, not a recommended method?

Our backups run nightly and they all use the option to not shut down, and use LVM snapshot if possible.

Again, thank you so much for the feedback, this is very helpful for us moving forward. Regards,

Right now, the only work-around is to not use the option to take a live snapshot when cloning a VM - this will ensure that a separate copy of the LV is made, which can itself by cloned.

chadwick89's picture
Submitted by chadwick89 on Thu, 05/03/2018 - 17:29

Thanks again for the info - we have modified our workflow so that we no longer clone live VMs. VMs will be in a shutdown state before initiating any cloning processes.

Regards,

Status: Fixed ยป Closed (fixed)

Automatically closed - issue fixed for 2 weeks with no activity.