Hi, I tried v7 very quickly with the backup-provider-dir-example storage. Some minor things I noticed below, IMO nothing that can't be fixed in follow-ups.
Note that I didn't look too closely how the backup provider API really works, so some of the following may just be consequences of the example backup-provider-dir-example plugin being an example plugin, in this case feel free to ignore: - naively triggering a one-off VM backup fails with "ERROR: cannot setup backup access without fleecing". I understand why and this is nothing critical, but I can imagine this could cause some confusion (e.g. if users don't immediately read the relevant ERROR line of the task log). Might be nice to either allow enabling fleecing also for one-off backups, or disable the "Backup Now" button for external storages. - I needed to install nbdfuse and modprobe nbd -- might be nice to do both automatically somehow (though packages of external backup plugins could probably add nbdfuse as a dependency?) - by accident I let the destination filesystem run full, which resulted in a stuck backup task and the following message being spammed to the journal: > Apr 02 16:46:34 pve-backupprovider kernel: block nbd0: Other side returned error (28) Maybe there is a way to abort the backup in such a case? - the VM backup wrote out the full 32G disk image including zeroes to the qcow2, but from a quick look at the source this is probably expected? Restoring this backup to e.g. lvm-thin gets rid of the zeroes again, though. - navigating to a VM backup on the backup-provider-dir-example storage and clicking on "Edit Notes" fails with > failed to stat 'PVE::Storage::Custom::BackupProviderDirExamplePlugin' (500) and the following message in the journal: > Apr 02 17:04:38 pve-backupprovider pvedaemon[960]: file_size_info: detected call without $file_format parameter _______________________________________________ pve-devel mailing list pve-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
