v0.32.1

• Introduce a new TRUST_PROXY env variable. Set this to true if you run zerobyte behind a reverse-proxy and want to avoid all client sharing the same rate limiting bucket.
• Protect app internal files to be overwritten by mistake during a custom target restore operation

v0.32.0

⚠️ Breaking Changes

• Dynamic secret references have been removed. env://... and file://... values are no longer resolved from repository or volume settings. It was replaced by startup-time provisioning instead: define those resources in a provisioning file and let Zerobyte resolve secret references during startup. This feature was removed because it is too difficult to secure properly in an multi-user environment.
• Before upgrading, replace any existing env://... or file://... values with the real secrets, or migrate those resources to file provisioning. Otherwise, affected connections may fail after the update. See example documentation
• If you didn't use dynamic secret references, nothing changes for you
• I have opened a discussion if anyone wants to ask questions or understand in more details why this had to be removed

Improvements

• Backup details now include a snapshot order toggle, so you can switch between oldest-first and newest-first views

Bug Fixes

• Snapshot and archive downloads now handle filenames with special characters more reliably
• Rejects include patterns that escape the selected volume root, preventing invalid backup configurations from being accepted
• Snapshot actions such as restore, download, delete, and tag now handle unusual snapshot IDs or refs that begin with - correctly
• Fixed permission handling for temporary credentials files, causing issues on some specific file systems

v0.31.0

New Features

• Zerobyte now includes a light theme, with a header toggle
• Backup ETA: running backups now show an estimated time remaining alongside progress details.

Bug Fixes

• Snapshot downloads are more reliable: dumping files from snapshots was fixed to avoid interrupted or incomplete downloads.
• Rclone-backed volumes no longer time out almost immediately because of an incorrect timeout conversion.
• Container timezone handling was fixed for certain timezones missing in the base alpine image

v0.30.0

New features

• OIDC Single sign-on: organization admins can now register OIDC providers and let users sign in through SSO.
• Invite-based organization access for SSO: admins can invite users with roles, and invited users are added to the organization on first sign-in.
• Organization switcher: users who belong to multiple organizations can now switch workspaces directly from the sidebar.
• Dedicated Administration area: instance-wide administration now has its own page, separate from regular settings.
• Advanced backup parameters: backup schedules now support custom Restic flags for advanced backup tuning.

Improvements

• Repository statistics now run in the background, which makes repository pages feel faster and reduces unnecessary compute on the server
• Repository, backup, and snapshot pages preload more data, so they show useful information sooner after navigation.
• The dashboard has been redesigned with small UI/UX improvements

Bug fixes

• Fixed single-file restore to a custom destination
• Live updates now use a shared connection, reducing duplicate refresh behavior across the app.

v0.29.3

• Replaced node-cron with a custom task scheduler implementation. It was causing server crashes on DST. Thanks @vutsalsinghal for the fix

v0.29.2

• Fix glob custom include patterns to expand relative to the volume root

v0.29.1

• Fix a schema issue that would skip the display of the backup progress if restic was not reporting any "current file"

v0.29.0

New features

• You can now download snapshot data directly from the restore page, including full snapshot downloads and single-item downloads.
• Snapshot exports support archive and file modes: folders/full selections download as .tar, while single files download as raw file streams.
• Repository details now include storage/compression stats (stored size, uncompressed size, space saved, ratio, snapshot count).
• Running backup progress is now cached and can be recovered after page reloads instead of starting from an empty progress view.


Improvements

• When restoring to a custom location, the selected folder for restoration will be put directly into the target folder. Without all the internal folder structure before

Bug fixes

• Fixed a backup path bug where selecting a subfolder with the same name as the volume could back up the whole volume root by mistake.
• Fixed Discord notifications so multiline messages are sent as a single message instead of one message per line.
• Fixed Gotify notifications for http:// servers by correctly setting DisableTLS=true.
• Improved timeout behavior so timed-out operations are reported as explicit errors instead of ambiguous failures.
• Increased rclone mount timeout handling to reduce premature failures on slower backends

v0.28.2

• Various UI and responsiveness fixes
• Keep sidebar last state when reloading page
• Fix local repository path on creation when using default path

v0.28.1

• Fix path migration for locally imported repos. If you see an imported local repo with a wrong path value, simply edit it and re-select the path
• Repositories are now editable
• When restoring a snapshot, a progress card is now displayed similar to the one when backing up
• Fix s3 url construction when user adds a trailing slash
• Fix an issue causing the backup schedule to stay even after deleting its volume or repository
• Fix file tree with more than 500 items at the root
• Allow usernames to contain dashes