[pve-devel] [PATCH docs 1/2] vzdump: add section about live-restore
d.whyte at proxmox.com
Thu Apr 22 12:30:06 CEST 2021
I just made some very minor suggestions inline, but overall it's good (i
had to search extra-hard for issues) :)
Reviewed-by: Dylan Whyte <d.whyte at proxmox.com>
On 4/22/21 10:25 AM, Stefan Reiter wrote:
> Signed-off-by: Stefan Reiter <s.reiter at proxmox.com>
> As discussed off-list, the vzdump section in general seems a bit misorganized,
> and is certainly missing some references to PBS. Dylan agreed to take a look at
> it, thanks again for that!
> vzdump.adoc | 28 ++++++++++++++++++++++++++++
> 1 file changed, 28 insertions(+)
> diff --git a/vzdump.adoc b/vzdump.adoc
> index 9453684..0577a97 100644
> --- a/vzdump.adoc
> +++ b/vzdump.adoc
> @@ -337,6 +337,34 @@ per configured storage, this can be done with:
> # pvesm set STORAGEID --bwlimit restore=KIBs
> +Restoring a large backup can take a long time, in which a guest is still
> +unavailable. For VM backups stored on a Proxmox Backup Server, this wait
> +time can be mitigated using the live-restore option.
> +Enabling live-restore via either the checkbox in the GUI or the `--live-restore`
> +property for `qmrestore` causes the VM to start immediately as the restore
s/property for/argument of/
> +begins. Data is copied in the background, chunks that the VM is actively
s/, chunks/, and chunks/
or to reformulate: "Data is copied in the background, prioritizing
chunks that the VM is actively accessing."
> +accessing are prioritized.
> +Note that this comes with two caveats:
> +* During live-restore, the VM will operate with limited disk read speeds, as
> + data has to be loaded from the backup server (once loaded it is immediately
s/once loaded/once loaded,/
> + available on the destination storage however, so accessing data twice only
> + incurs the penalty the first time). Write speeds are largely unaffected.
> +* If the live-restore fails for any reason, the VM will be left in an
> + unspecified state - that is, not all data might have been copied from the
> + backup, and it is _most likely_ not possible to keep any data that was written
> + during the failed restore operation.
> +This mode of operation is especially useful for large VMs, where only a small
> +amount of data is required for initial operation, e.g. web servers - once the OS
> +and necessary services are started, the VM is already operational, while the
s/are started/have been started/
> +background task keeps copying seldomly used data.
More information about the pve-devel