Feedback for Proxmox Backup Server

[reto]

I've setup two Proxmox Backup Servers, one at my parents house and one at my own place (running on Synology VMs). So far I'm quite happy. I'm planning to describe my Synology Docker Setup in a separate post.

I made some notes about points I noticed during the setup. Proxmox Team: If you like I can create bug reports for these points, or just leave them here in this post .


Documentation

• Backup Group / Backup ID: I think I got the hang of it, but It would be nice to see some further explanation or ideas how to organize backups. And what side-effects are there? I.e. as far as I can understand, the previous Snapshot is downloaded to calculate what delta needs to be uploaded. So it's kinda important to alway selected the same subset of data within a Backup Group. What are the allowed chars for a backup id?
• Tokens: Tokens cannot have more rights than their user, i.e. if a User has no rights, and you give Backup Rights to token of this user, you will actually no rights using the token. This makes sense if you think about it. Maybe even add a warning to the UI?
• Client: is --exclude relative to the backup root? Or absolute? I assume the former.
• Partial backups: What happens when a backup is aborted after 90% completion? Answer: the chunks are there, but will be reuploaded regardless. I used Excludes to split up large 'frist' backups with excludes.

Feature Requests:

• Better handling of aborted backups (especially initial backups): I had a slow link and had to retransmit over 24h of data after an aborted backup (short network interrupted). It would be nice, if you create 'partial' snapshots for aborted backups, maybe only for the first backup. This snapshots would be used for retries. Probably not so easy I know.
• Limit uploads: Regarding the points above: maybe provide an option: "--complete-backup-after-gb-uploaded NNN" which will successfully complete a backup after uploading the specified amount of data. This would make setting up large backups over slow links easier.
• Progress bar: It would be nice to see some progress during large backups. i.e. one printout per 10 minute, or per 1 GB or whatever.
• Generate ENV Variables for Data Stores: have small widget on the UI, on the view Datastore to 'Copy the Registry ENV Variable' to the clipboard (or just show them). i.e. for the current user or for a selectable user/token.

Small note: 1000€ per year is too steep for my (hobby) setup, but I see that you cannot reduce the price for us hobbiest without loosing sales for business customers. But I'd love to support you without access to enterprise repos (or maybe just disable the popup after login? ). For a lower nominal fee?

Anyway, thank you very much for PBS!

 

[fabian]

I've setup two Proxmox Backup Servers, one at my parents house and one at my own place (running on Synology VMs). So far I'm quite happy. I'm planning to describe my Synology Docker Setup in a separate post.

I made some notes about points I noticed during the setup. Proxmox Team: If you like I can create bug reports for these points, or just leave them here in this post .


Documentation

• Backup Group / Backup ID: I think I got the hang of it, but It would be nice to see some further explanation or ideas how to organize backups. And what side-effects are there? I.e. as far as I can understand, the previous Snapshot is downloaded to calculate what delta needs to be uploaded. So it's kinda important to alway selected the same subset of data within a Backup Group. What are the allowed chars for a backup id?

a backup group consists of a type (vm/ct/host) and an ID. together they represent the entity that backups operate on - e.g., creating a new snapshot, pruning existing ones, incremental comparison to the last existing snapshot within the group. for PVE-initiated backups, the ID will always be the guest ID, so you'll have a single backup group per guest (unless you are using multiple datastores configured as multiple storages). the regular PBS client will use the hostname as ID by default for 'host' type backups.

https://pbs.proxmox.com/docs/terminology.html#backup-group

• Tokens: Tokens cannot have more rights than their user, i.e. if a User has no rights, and you give Backup Rights to token of this user, you will actually no rights using the token. This makes sense if you think about it. Maybe even add a warning to the UI?

documented

https://pbs.proxmox.com/docs/user-management.html#api-token-permissions

• Client: is --exclude relative to the backup root? Or absolute? I assume the former.

it works just like the patterns from files (think of the CLI --exclude like a top-level .pxar-exclude file)
https://pbs.proxmox.com/docs/backup-client.html#excluding-files-directories-from-a-backup

• Partial backups: What happens when a backup is aborted after 90% completion? Answer: the chunks are there, but will be reuploaded regardless. I used Excludes to split up large 'frist' backups with excludes.

the chunks remain until the (one of the) next GC collects them, the exact logic is a bit complicated, but usually the sequence

- start backup
- interrupt backup
- start another backup with same source and target

should be able to pick up all the chunks from the first run, but likely only server-side (so the chunks would still be generated and uploaded, just discarded since they already exist).

Feature Requests:

• Better handling of aborted backups (especially initial backups): I had a slow link and had to retransmit over 24h of data after an aborted backup (short network interrupted). It would be nice, if you create 'partial' snapshots for aborted backups, maybe only for the first backup. This snapshots would be used for retries. Probably not so easy I know.

something like that could be done - but it complicates the UX quite a lot (how to display those 'partial' snapshots? which operations are okay to do, which aren't? ..)

usually what people do if they have a lot of data that needs to be backed up at the start to bootstrap is to locally install PBS, do the intiial backup there, and then either sneakernet the actual datastore or sync over the net using the built-in sync mechanism (that handles interruptions a lot more gracefully than the backup mechanism, since it is pull and can just pick up again).

• Limit uploads: Regarding the points above: maybe provide an option: "--complete-backup-after-gb-uploaded NNN" which will successfully complete a backup after uploading the specified amount of data. This would make setting up large backups over slow links easier.

see above

• Progress bar: It would be nice to see some progress during large backups. i.e. one printout per 10 minute, or per 1 GB or whatever.

yeah, better progress reporting for host/ct backups and restores would be a nice UX improvement.

• Generate ENV Variables for Data Stores: have small widget on the UI, on the view Datastore to 'Copy the Registry ENV Variable' to the clipboard (or just show them). i.e. for the current user or for a selectable user/token.

that is a nice suggestion! I'd also like to see a 'generate token + ACL' wizard where this would nicely fit in

Small note: 1000€ per year is too steep for my (hobby) setup, but I see that you cannot reduce the price for us hobbiest without loosing sales for business customers. But I'd love to support you without access to enterprise repos (or maybe just disable the popup after login? ). For a lower nominal fee?

no such plans atm - but you can always support the devs by getting community subscriptions where you can afford them, and you do get the full range of features for your hobbyist setup in either case anyway.

Anyway, thank you very much for PBS!

thanks for your feedback!