> On March 19, 2020 1:05 PM Thomas Lamprecht <t.lampre...@proxmox.com> wrote: > > > On 3/19/20 11:51 AM, Fabian Ebner wrote: > > Hi, > > this does fix an issue when the receiving side has the most recent > > snapshot, but not the 'old_snap' one. And of course testing for 'last_snap' > > is correct, since that one is the relevant one for the incremental sync. > > With that: > > > > Reviewed-By: Fabian Ebner <f.eb...@proxmox.com> > > Tested-By: Fabian Ebner <f.eb...@proxmox.com> > > With that applied, thanks to all. > Wolfgang you missed your signed-off-by (FYI).
Will do the next time. > > > > Some ideas for further improvements: > > * If on the destination there are older snapshots but not most recent one, > > pve-zsync tries to do a full sync and fails with: "destination has > > snapshots". One could get the list of snapshots from the destination, the > > list of snapshots from the source and use the most recent common one as the > > starting point for an incremental sync (and fail if both sides do have > > snapshots but no match). > > * Knowing the list of snapshots for the destination could also be used to > > prevent issuing a useless remote 'zfs destroy' when the snapshot to be > > deleted does not exist for the destination. > > Sounds reasonable, IMO. I've already thought about it, but I wouldn't do it automatically. More Like this pve-zsync cleanup: Removes all snapshots with rep_ on behalf of the source host that are not available on the target. pve-zsync fixSync: searches for the last common synchronization point and removes all subsequent ones at the destination One problem with automatic resolution is that multiple replications can accidentally delete snapshots. > > > _______________________________________________ > pve-devel mailing list > pve-devel@pve.proxmox.com > https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel _______________________________________________ pve-devel mailing list pve-devel@pve.proxmox.com https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
