[1.33.0-rc.2] - 2025-12-01
Fixed
- AutoSnapshot: Fixed "Keep" property (regression introduced in 1.32.0)
- Fixed bug where clone no longer works with other min-io size than 512
- Fixed bug where "r c" fails with disabled auto-min-io on a 4k block-sized SP
- Clone: cleanup source snapshot if cloning failed
- Clone: mark failed volumes with DELETE flags to allow to delete them afterward
- Fixed --diskless-on-remaining not working with COMBINED nodes
- Fixed unnecessary restart of local DRBD peer right after the resource was created
- ToggleDisk: Added missing free-space check
- Clone: Added missing free-space check
- Fixed very rare AccessToDeletedNode during a node-update from the satellite
- Fixed possible NPE in DrbdLayer.hasMetaData
- If one resource fails during calculating gross/net size, other resources should still be processed
- Check resource size on creation and resize
- Run autohelper.manage tasks after node fullsync (before only on node rc, c, d)
- Autotiebreaker: do not delete tiebereaker if node is not online/fullsynced
- Cleanup clone task on satellite reconnect
- Clone: Fixed race conditions when starting multiple clones at once
[1.33.0-rc.1] - 2025-11-11
Added
- physical-storage-api: allow passing pv/vg/lv/zpool create arguments
- Encryption: Added systemd-creds support for masterpassphrase.
- DRBD: Added Linstor/Drbd/auto-block-size
- Backup: allow shipment of already existing snapshots
- Node,evacuate: Added --target and --do-not-target options
Changed
- Removed keep_res regular expression pattern from options
- Instead of caching if a DRBD .res file exists, actually check if the file exists
- Do not delete known Linstor drbd res files, LS_KEEP_RES is now obsolete
- Increase max packet size 16M -> 64M
- Increased timeout for tar during sos-report to an hour
- Removed flywaydb dependency and do migrations (as with other "db"s) ourselves
- ReconnectorTask: changed logging levels and added more logging
- API: do not store error reports if required remote parameters are missing
- Node evacuate: By default node evacuate also evacuates existing snapshots
Fixed
- Attempt to fix race-condition "Received unknown storage resource from satellite"
- Prometheus: fix using peerId as peer address in api call duration statistics (GitHub #466)
- H2 db should no longer prematurely shutdown
- DRBD: recreate ext-metadata after backing disk needed to be re-created
- Fixed a bug where a failed clone leads to AccessToDeletedData bug
- Cache: Properly flush to origin device when suspending
- Fixed toggle-disk bug when evacuation target has already diskless resource
- Fixed possible NPE when a satellite updates a storage pool (capacity, supports snapshots, etc)
NOTE
With this release we also announce that we will remove etcd database support with Linstor 1.34.0, until then you will have time to migrate to (SQL or k8s-CRD) databases.