ProxSave v0.12.8
🧠 Safer network rollback, clearer cloud handling, and improved restore interaction
This release focuses on restore-time safety, clarity, and usability, with major improvements to network rollback handling and diagnostics, safer interactive prompts with timeouts, clearer cloud path behavior, and refined cluster and datastore logic. It also completes the transition to fully English documentation and comments.
- Network rollback safety hardening: Improved rollback handling to remain armed until explicitly committed, prevent late commits, detect already-running rollbacks, and provide clearer state-aware error reporting.
- Rollback diagnostics and visibility: Added detailed rollback state logging, current and restored IP reporting, rollback countdown displays, and enriched abort information to make network recovery status explicit after aborts.
- Interactive safety with countdown prompts: Introduced visible 90-second countdowns for critical restore prompts (network apply and smart
/etc/fstabmerge), defaulting to safe actions when unattended or aborted. - Abort-aware restore workflows: Enhanced restore logic to better handle user-initiated aborts, propagate abort context, and surface rollback status and deadlines consistently in logs and UI.
- Cloud log path normalization: Added consistent cloud log path normalization via
cloudLogBase, ensuring correct remote prefix handling and predictable log cleanup behavior. - Clearer cloud configuration guidance: Updated documentation to clarify cloud path formats and provide concrete examples, along with expanded troubleshooting guidance for network rollback and SSH/IP changes.
- Refined cluster detection: Improved cluster detection to require both multiple node directories and valid corosync configuration, avoiding false positives on standalone nodes.
- Expanded BACKUP_CLUSTER_CONFIG behavior: The flag now controls both cluster configuration backup and runtime cluster command collection (
pvecm, HA status), with updated docs and tests. - Datastore ownership robustness: Updated ownership handling to safely skip chown/chmod when non-root and treat permission-related errors as non-fatal, improving behavior on restricted or CI environments.
- Improved rclone cloud scan timeouts: Cloud restore and decrypt workflows now cap scan duration using
RCLONE_TIMEOUT_CONNECTION, with clearer error messages, debug logs, and updated documentation. - Documentation and language consistency: Translated comments, documentation, and user-facing strings from Italian to English across the codebase, with no functional changes.
Overall: safer unattended restores, clearer rollback visibility, more predictable cloud behavior, refined cluster handling, and improved operator confidence during recovery scenarios.