linstor-server 1.31.0 release

[Rene Peinthor]

Hi!

Here is finally the final release of linstor 1.31.0, we noticed some design
deficits with the
new replication states, so had to rework them a bit, so it took a bit
longer after the last RC.
We also fixed a few smaller bugs we found while testing.

Some highlights of this release:
  * Fixed snapshot rollback
    Snapshot rollback only worked if the resource/node the snapshot was
made, did never
    get deleted or moved to another node. As the new node would never have
the origin
    snapshot. We came up with a node resync idea and for ZFS with a
renaming and refcounting
    method to make sure snapshot rollback always works.
  * Auto-quorum reworked
    The auto-quorum property was very special and hard to understand as it
didn't follow
    the Linstor principle of inheriting properties. It usually copied its
value into the on-no-quorum
    property and you had to change each of them, if you wanted to update
all of them.
    Now the property is gone and the behaviour will only be controlled by
the DRBD option
    quorum.
  * DRBD replication states are now sent to the linstor-controller
    These states are used in the balance resource task and also shown in
the client.

## [1.31.0] - 2025-04-08

### Changed

- Improved TCPConnector logging and only reconnect in client mode
- physical-storage-api: allow VDO with lvm-thin
- ETCD: Enabled DrbdOptions/auto-diskful-allow-cleanup on ctrl unless it
was already set
- ReplicationState design wasn't considering replication states per volume
connection.
  This has been reworked and now a map is used and returned to the client.

### Fixed

- Weird access to deleted resource bug on drbd down
- CreateNetIf didn't close current peer connection
- UpdateSpaceInfoTask did never commit() its allocation size updates
- Fix deleting of a drbd resource if it was already down

## [1.31.0-rc.2] - 2025-03-21

### Fixed

- Add missing migrations (etcd/crd) for ZFS rename
- Only satellite zfs migrate nodes that have a ZFS pool

## [1.31.0-rc.1] - 2025-03-19

### Added

- Notify controller about replication state changes in DRBD
- Notify controller about done percentage changes in DRBD
- Encryption-API: added a GET endpoint to ask for the master passphrase
status

### Changed

- BalanceResource: Ignore resources which are in a non-valid replication
state
- sos-report: include backup ship queue
- sos-report: syslog will only be gathered if messages was not available
- Clone: check that a resource name for the cloned resource is set
- Clone: fail clone if the source resource isn't UpToDate or has skip-disk
- Memory: Limit satellite to 2G max memory (-Xmx2G) and controller to 8G
- JVM: will now crash on OutOfMemoryError
- Limit reactor worker threads to 16
- DRBD .res file now contains the time the file was generated
- BalanceResource: Improved selection when deleting excess resource
- AutoDiskfulTask: Improved selection when deleting excess resource
- Tiebreaker: Improved decision making if a deleting resource should be
kept as tiebreaker
- Write a full documented linstor.toml and remove the linstor.toml-example
- Snapshot rollback now does internally a resource delete + snapshot restore
- ZFS resource can now be deleted while having ZFS snapshots (ZFS volumes
will be renamed)
- ZFS snapshots can now be deleted while having ZFS clones (ZFS snapshots
will be renamed)
- Auto-quorum: Reworked auto-quorum property behaviour, property was
removed and auto-quorum is now managed through
               the drbd option quorum. on-no-quorum properties will be
inherited.
- BackupShip: start next queued shipment immediately when sending fails

### Fixed

- Reconnect: Fixed rare race condition in DRBD version check
- Resource delete: Fixed rare race condition where a delayed DRBD event
causes "resource not found" ErrorReports
- Fixed possible UUID mismatch after deletion of last snapshot failed to
properly clean up satellite
- Fixed "resource not found" bug when restoring from a snapshot when the
source-resource was already deleted
- Satellite: Fixed rare "attempt to replace active transMgr" error
- SATELLITES_CAPACITY table NODE_NAME size was too short
- BackupShip: queued backups are now removed if the target cluster goes
offline
- BackupShip: do not use snapshots that are deleting as base for an
incremental backup
- BackupShip: remove snapshots from queued backups that have them as base
when deleting them

https://pkg.linbit.com//downloads/linstor/linstor-server-1.31.0.tar.gz


Linstor PPA:
https://launchpad.net/~linbit/+archive/ubuntu/linbit-drbd9-stack

Cheers,
Rene
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-announce/attachments/20250408/894d06ca/attachment.htm>