Newer
■
■
■
■
■
■
■
■
■
■
Older
104 lines
|
4.749kb
| 1 |
# PGS - Changelog |
|
| 2 | ||
| 3 |
## [1.5] - 2026-03-07 |
|
| 4 | ||
| 5 |
### Added |
|
| 6 |
- Added `pgs cleanup` to scan image storages for orphan/stale `vm-*-state-suspend-YYYY-MM-DD.raw` volumes and remove them safely |
|
| 7 | ||
| 8 |
### Fixed |
|
| 9 |
- Stopped VMs are no longer classified as "already suspended to disk" from config flags alone; `bin/pgs` now requires `lock: suspended`, `vmstate:`, and a resolvable backing saved-state volume |
|
| 10 |
- Added cleanup for inconsistent suspend artifacts on stopped VMs, including stale suspend locks, stale `vmstate:` metadata, and orphaned saved-state volumes on storage |
|
| 11 |
- `pgs suspend` now runs suspend-artifact cleanup as a preflight, reducing same-day collisions with stale `state-suspend` volumes |
|
| 12 |
- Cleanup explicitly ignores `vm-*-state-cp*.raw` checkpoint files and only targets `vm-*-state-suspend-YYYY-MM-DD.raw` |
|
| 13 |
- Repeated `pgs suspend` runs now merge with the existing state file instead of discarding prior `to_resume` intent |
|
| 14 |
- State now records `vm_details.suspend_volume` and `vm_details.suspend_file_date`, and `resume` skips auto-restore when a VM's suspend artifact changed after the state was saved |
|
| 15 |
- Suspend-artifact cleanup now scans only local `dir` storages; remote storages such as NFS are skipped so planned maintenance cannot block in kernel I/O wait on a stale mount [ISSUE-2026-002] |
|
| 16 | ||
| 17 |
## [1.4] - 2026-03-06 |
|
| 18 | ||
| 19 |
### Changed |
|
| 20 |
- Standardized install layout around `xdev` paths for uninstall, documentation, and runtime state |
|
| 21 |
- Added dedicated `scripts/install.sh` and `scripts/uninstall.sh` and reduced `setup.sh` to a local/remote wrapper |
|
| 22 |
- Updated `bin/pgs` to migrate legacy state from `/var/lib/pve-manager/pgs-state.json` to `/var/lib/xdev/pve-guests-state/pgs-state.json` |
|
| 23 |
- Promoted `bin/pgs` as the canonical executable and removed the duplicate top-level `pgs` file |
|
| 24 |
- Marked `systemd/` artifacts as legacy reference material instead of active install targets |
|
| 25 | ||
| 26 |
### Fixed |
|
| 27 |
- Fixed documentation to reflect the current manual workflow and the standardized host layout |
|
| 28 | ||
| 29 |
## [1.2] - 2026-03-05 |
|
| 30 | ||
| 31 |
### Added |
|
| 32 |
- LXC container (CT) support: graceful shutdown before maintenance, auto-start after maintenance |
|
| 33 |
- New `ct_to_start` array in state JSON for CT restoration |
|
| 34 |
- `load_ct_info()` function using single `pct list` call |
|
| 35 |
- `shutdown_ct()` function with 120s timeout for graceful shutdown |
|
| 36 |
- `start_ct()` function for post-maintenance startup |
|
| 37 |
- TODO placeholder for critical VM/CT migration support |
|
| 38 | ||
| 39 |
### Changed |
|
| 40 |
- State file now includes `ct_to_start` array |
|
| 41 |
- Suspend operation processes VMs then CTs |
|
| 42 |
- Resume operation resumes VMs then starts CTs |
|
| 43 | ||
| 44 |
### Fixed |
|
| 45 |
- Fixed `pct list` column parsing (Status/Lock/Name column order) |
|
| 46 |
- Handle empty Lock column in `pct list` output |
|
| 47 | ||
| 48 |
## [1.1] - 2026-03-05 |
|
| 49 | ||
| 50 |
### Fixed |
|
| 51 |
- Fixed `load_state()` outputting log messages to stdout, corrupting JSON parsing |
|
| 52 |
- Fixed empty arrays in JSON state file (was generating `[""]` instead of `[]`) |
|
| 53 |
- Fixed paused VMs being treated as "running" - now properly detects `paused` status |
|
| 54 | ||
| 55 |
### Changed |
|
| 56 |
- Optimized VM info loading: single `qm list` call instead of per-VM calls |
|
| 57 |
- Optimized suspend lock detection: read config files directly, no extra `qm` calls |
|
| 58 |
- Optimized status checking: only verify actual status for "running" VMs, rest trust `qm list` |
|
| 59 |
- Reduced scan time from ~180 seconds to ~2.5 seconds for 30+ VMs |
|
| 60 | ||
| 61 |
### Added |
|
| 62 |
- Proper systemd service setup for manual suspend before maintenance |
|
| 63 |
- Proper systemd service setup for manual resume after maintenance |
|
| 64 |
- Better handling of paused VMs: suspend to disk but don't auto-resume |
|
| 65 |
- Comprehensive journal logging with severity levels (INFO, WARNING, ERROR, SUCCESS) |
|
| 66 |
- Dry-run mode for testing without effects |
|
| 67 | ||
| 68 |
## [1.0] - 2026-03-05 |
|
| 69 | ||
| 70 |
### Initial Release |
|
| 71 |
- Basic suspend/resume functionality |
|
| 72 |
- State file preservation |
|
| 73 |
- Manual testing scripts |
|
| 74 | ||
| 75 |
--- |
|
| 76 | ||
| 77 |
## Performance Improvements |
|
| 78 | ||
| 79 |
| Operation | v1.0 | v1.1 | Improvement | |
|
| 80 |
|-----------|------|------|-------------| |
|
| 81 |
| Scan 30 VMs | ~180s | ~2.5s | **72x faster** | |
|
| 82 |
| System calls | Per-VM qm calls | Single qm list + file I/O | **Drastically reduced** | |
|
| 83 | ||
| 84 |
## Known Limitations |
|
| 85 | ||
| 86 |
- Requires passwordless SSH for cluster-wide operations |
|
| 87 |
- No critical VM/CT migration support yet (TODO) |
|
| 88 | ||
| 89 |
## Testing |
|
| 90 | ||
| 91 |
Tested on: |
|
| 92 |
- Proxmox VE 8.x with 30+ VMs and CTs |
|
| 93 |
- Mixed VM configurations (4GB-16GB RAM) |
|
| 94 |
- LXC containers with running services |
|
| 95 |
- Storage: local-dir, NFS mount points |
|
| 96 |
- Planned reboot validation on `baobab`: shutdown-to-ICMP-loss improved from ~106s to ~15s once NFS over Thunderbolt stopped losing transport before unmount |
|
| 97 | ||
| 98 |
## Future Enhancements |
|
| 99 | ||
| 100 |
- [ ] Support for LXC container shutdown |
|
| 101 |
- [ ] Configurable exclusion list for VMs |
|
| 102 |
- [ ] Metrics/performance monitoring |
|
| 103 |
- [ ] Multi-node coordination for cluster-wide operations |
|
| 104 |
- [ ] Backup integration for backup snapshots before suspend |