Proxmox File Level Restore Portal (Open Source) with snapshot timeline and guest restore

[HappyHog]

Proxmox's file-restore feature (browse/download individual files out of a PBS backup via the helper-VM) already does the hard part — it's just missing a way to scrub across snapshots instead of picking one at a time from a flat dropdown. I built a small companion app on top of it, in the spirit of Synology Active Backup for Business's restore browser.



What it does:
- A drag-to-scrub timeline across every snapshot of a guest, instead of a flat list
- Browse/download files or whole directories (.zip/.tar.gz/.tar.zst) out of any snapshot
- Restore a file or directory straight back into the running guest via qemu-guest-agent (including a faster direct-network-transfer path for large content)
- Logs in with your own PVE account (including OIDC/SSO realms as of the latest release) — no shared service token, no direct PBS access; PVE's own permissions decide what you can see
- Multiple PBS storages/namespaces, colour themes, LXC/Docker deployment

It doesn't touch Proxmox itself — it's a separate app that talks to the same PVE API the web UI does.

Built for a single-admin homelab scale (my own 2-node cluster), not tested at anything bigger. AGPLv3, source and setup instructions here:

https://github.com/treycentric/pve-flr-portal

Latest release: v1.5.0 — https://github.com/treycentric/pve-flr-portal/releases/tag/v1.5.0

Happy to answer questions or take issues/PRs if anyone tries it out.

 

[dcsapak]

IMO the ui looks nice and i get why this is a feature that some want. but there is currently a big caveat: for each pbs snapshot that is mounted there will be an extra helper vm that consumes memory and cpu resources. these will be auto cleanup after a timeout, so if you open too many of those the host will likely oom kill some processes (including maybe an unrelated vms)

 

[HappyHog]

IMO the ui looks nice and i get why this is a feature that some want. but there is currently a big caveat: for each pbs snapshot that is mounted there will be an extra helper vm that consumes memory and cpu resources. these will be auto cleanup after a timeout, so if you open too many of those the host will likely oom kill some processes (including maybe an unrelated vms)

Thanks for the feedback. I have added a throttling config setting (defaulted to 4) and pooling and queuing of API calls to try and limit runaway spin ups of helper VMs.

 

[HappyHog]

Proxmox's file-restore feature (browse/download individual files out of a PBS backup via the helper-VM) already does the hard part — it's just missing a way to scrub across snapshots instead of picking one at a time from a flat dropdown. I built a small companion app on top of it, in the spirit of Synology Active Backup for Business's restore browser.

View attachment 100106

What it does:
- A drag-to-scrub timeline across every snapshot of a guest, instead of a flat list
- Browse/download files or whole directories (.zip/.tar.gz/.tar.zst) out of any snapshot
- Restore a file or directory straight back into the running guest via qemu-guest-agent (including a faster direct-network-transfer path for large content)
- Logs in with your own PVE account (including OIDC/SSO realms as of the latest release) — no shared service token, no direct PBS access; PVE's own permissions decide what you can see
- Multiple PBS storages/namespaces, colour themes, LXC/Docker deployment

It doesn't touch Proxmox itself — it's a separate app that talks to the same PVE API the web UI does.

Built for a single-admin homelab scale (my own 2-node cluster), not tested at anything bigger. AGPLv3, source and setup instructions here:

https://github.com/treycentric/pve-flr-portal

Latest release: v1.5.0 — https://github.com/treycentric/pve-flr-portal/releases/tag/v1.5.0

Happy to answer questions or take issues/PRs if anyone tries it out.

Renamed to "File Level Restore Portal for Proxmox Backup Server", based on Media Kit guidance. Latest version is now v1.6.0.

 

[scyto]

thanks for the continued improvement, saw your post on reddit on this, I will get to testing this weekend
(first task was improving my cephFS backup to pbs strategy, thats nearly done)