On Wed, Aug 27, 2025 at 13:04:05 -0000, pjbo...@systemdatabase.com wrote:
> Hello Peter:
>
> Thank you for your reply.
>
> I've found the page difficult to follow. There's 13 steps listed, but I
> found the numbering and heading confusing. I've followed 10 of the 13 steps.
Yes, the numbering was broken due to a mistake in the source document.
I've posted a patch to fix it.
>
> After following these 10 steps, I'm asking about purge of files. Have I
> followed the steps incorrectly?
>
> 1. (FOLLOWED) Start the guest:
> 2. (FOLLOWED) Enumerate the disk(s) in use:
> 3. (FOLLOWED) Begin the backup:
> 4. (FOLLOWED) Check the job status ("None" means the job has likely
> completed):
> 5. (FOLLOWED) Check the completed job status:
> 6. (FOLLOWED) Now we see the copy of the backup:
Wait so you follwed the steps bot for the backup with new libvirt and
old libvirt?
The first set of steps is sufficient if you have a new enough libvirt
and the older steps describe just steps to achieve identical result
albeit with more complex approach.
So if 'virsh backup-begin' works for you, you are supposed to ignore the
rest.
>
> Full backup with older libvirt versions
> 1. (considered optional) Start with a guest with a single disk image,
> base.raw, which is where the live QEMU is pointing at, and recording the
> guest writes:
> 2. (considered optional) List the current block device(s) in use:
This step is there so that you can compare the before and after as you
need to make sure to delete the correct file
> 3. (considered optional) Create the live "external disk snapshot" (or more
> correctly, "an overlay"):
this is NOT optional if you are doing the "fallback" method for old
libvirt versions
>
> 1. (FOLLOWED) Now, take a backup the original image, base.raw, to a different
> location using cp or rsync:
> 2. (FOLLOWED) Enumerate the current block device(s) in use, again. Notice
> that the current disk image in use is the above-created overlay,
> overlay1.qcow2:
> 3. (FOLLOWED) Once the backup of the original image completes, now perform
> the "active block-commit" to live-merge the contents of overlay1.qcow2 into
> base.raw and pivot the live QEMU back to the original:
> 4. (FOLLOWED) After the above operation completes, again list the current
> block device(s) in use. And notice that the live QEMU is now writing to the
> original base image:
And all of this makes no sense after you skip the step that I've replied
above.
With 'virsh backup' there's nothing to clean up because the only result
is the actual backup file as no intermediates were generated.
The blockcommit operation ought to be safe but don't delete anything at
this point since you skipped the step that creates the extra overlays.
