Release Notes


Release Notes V8 build is 11456 (08 August 2017)

Download here

Core

Fixed issue with hang-up on service stopping. On service stop it could get stuck with repeating messages in log: "Waiting for 5 seconds for cleanup to complete (XX refs)...". Messages are repeated without changing the number of references and without actual stop of execution.

Synchronous Replication

Fixed issue with incorrect processing of time-outs for commands on synchronization channel. Error in processing of timed-out commands could lead to crash of the service.

ODX performance optimizations.


Release Notes V8 build is 11404 (07 July 2017)

Cache

Fixed processing of TRIM/UNMAP command. When cache memory allocation was still in progress, but virtual devices have already been set to the functional state, service could crash on receiving TRIM/UNMAP command.

Synchronous Replication

Fixed issue with "Synchronization request will be refused" state in logs. In some cases, devices with synchronous replication could not start synchronization after a server reboot, and the message "Synchronization request will be refused" appeared in service logs.

Known issue

LSFS-based devices with synchronous replication could lose synchronization when disk load was growing significantly due to replica synchronization or cache flushing.


Release Notes V8 build is 11156 (24 May 2017)
Cloud Replicator for VTL
Amazon S3 integration libraries updated.

Licensing

New license type added: Cloud VTL. Please check web site for more information.

Synchronous Replication

Fix for processing of ODX commands when ODX support is disabled in configuration file. In case when ODX is disabled in configuration file and client sends ODX commands, service could crash on processing command errors.


Release Notes V8 build is 11071 (12 May 2017)

Virtual Tape

Functionality has been added for replicating Virtual Tape files to cloud. Amazon S3 with migration to Glacier are now supported.
Please check the website for more information on this feature.

StarWindX PowerShell module

Extend command has been added for HA devices.

LSFS

Fixed issue: saving metadata in "close device" procedure. On large amounts of user data in some cases, this error could lead to issues with mounting LSFS device on the following "open device" procedure.
Fixed crash: service had both LSFS device and its snapshot mounted and the user removed source LSFS device.

Free License

Fixed issue: service was blocking iSCSI connections when the trial period has expired.


Release Notes V8 build is 10927 (19 April 2017)

StarWindX (PowerShell module and COM-object)

Fix for operating on Core versions of Windows. PowerShell commandlets generated errors on Core 64-bit versions of Windows OS.


Release Notes V8 build is 10914 (13 April 2017)

Cache

Improved procedure of memory allocation for cache on device creation. Now this process executes when device is already created and functioning, so adding device with large cache completes fast.

Management Console

Fixed issue with setting NUMA node for new device and target. This parameter remained not set. This error was introduced in build 10768.


Release Notes V8 build is 10833 (3 April 2017)

Cache

Fixed error in processing read requests with a write-back cache. In case of a large cache volume (gigabytes), read operation started after a long write load period and could be processed with significant delay.

This error has appeared in build 10547.

Virtual Tape

Fixed error in processing a virtual tape inserted in a Tape Library with several virtual tape drives. The service could crash on startup if rebooted with a tape inside a virtual tape drive.


Release Notes V8 build is 10811 (27 March 2017)

Synchronous replication

Compatibility with old device configurations starting from V6 without ALUA settings has been added. In previous builds, service crashed when reading such device configuration strings.

Detection of network connection loss has been fixed. After the sequence of disconnect and connect operations executed by HA node on sync channel, synchronization channel status could be incorrectly displayed as functional when the connection was lost.

iSER protocol

Fixed issue: when a user disabled network adapter after starting the service and enabling iSER protocol, the service could crash. Now, the service runs correctly during this operation.


Release Notes V8 build is 10799 (14 March 2017)

Management Console

Fixed issue with non-default target name for a new device.
When a user created a new device and set target name other than the default, this setting was ignored and target name was set back to default value.

Fixed issue with using replication manager on devices that are created on StarWind Virtual SAN V6.
When a new synchronous replica was being added to existing device that has been created in version 6, the operation resulted in error.

Fixed issue with adding a replica to a device with size greater than 16TB.
When a new replica was being added to a device with size greater than 16TB, the operation resulted in error with message "There is not enough space on the disk".

LSFS

Added implementation of XCOPY command for LSFS-based devices. This command is part of VAAI command set.

Changed operation of LSFS device with write-back cache in low disk space circumstances. Disk write operations are throttled to speed up journal defragmentation process.

Synchronous replication

Synchronization using snapshots (for LSFS devices) has been changed to take snapshots synchronously.
Preparation before snapshot has been added into fast snapshot synchronization routine. It reduces time into "start synchronization guard" scope and prevents fail of the network connection between partners due to timeout.

Fixed issue: in the case of storage fail on one node, this node is moved to offline mode. This process includes flushing cache to disk. When the cache has a large size, this operation takes significant time. In previous versions, the node has been moved to offline mode after flush cache operation had been finished, in a result client could send requests to the non-functional node during this period. Now, this behavior is changed and the node is switched to offline mode before flushing cache to disk.

iSER protocol

Fixed issue: HA devices could lose synchronization under high load when iSER protocol has been used for sync channel.


Release Notes V8 build is 10695 (24 February 2017)

Core

Fixed bug: StarWind Virtual SAN application got hanged on startup after Windows updates were installed and some Windows features/components were in the "Waiting for restart" state

Synchronous replication

Compatibility with V6 device headers has returned. V6 used device headers in binary format and starting from V8 StarWind applies XML format for swdsk device header files. HA device automatically converts V6 format into V8 on startup.
We recommend checking with StarWind Support and Knowledge base before upgrading old versions.


Release Notes V8 build is 10547 (26 January 2017)

Core

Extended support for NUMA architecture: option to set NUMA node for Virtual Device and Target, automatic distribution of devices and targets on NUMA nodes. Added support for ODX commands (Windows Offloaded Data Transfer). ODX performs certain data transfer operations on storage side without moving data through the client machine. Performance optimizations for VAAI WRITE_SAME command.

iSER Protocol

iSER protocol support added as experimental functionality. Network cards with RoCE v2, RoCE v1 RDMA protocols support are required for the protocol to work. Use Server - Configuration - Network section in Management Console to enable iSER support.
iSER can be used for client connections and for synchronous replication. Mixed scenarios are supported: iSCSI access for client connections and iSER protocol on HA synchronization channel.
Please join our forums for discussion.

VMware Virtual Volumes

Added support of the HA devices to the StarWind VASA Provider as experimental functionality. HA devices presented as HA VVols. VASA Provider allows administrators to configure StarWind Virtual SAN storage resources as storage containers for virtual machines and enables storage policy-based management (SPBM) of virtual machines. Use WMI to configure VASA Provider. Documentation is available on the Web site.

Cache

Statistics for cache operations: cache hits rate, cache usage percent. This statistics is being accumulated starting from device creation or service start and can be reset for checking the exact time period.

Flash Cache

Added performance optimizations for flash cache module.

Synchronous Replication

Traffic priority management for sync channel has been reworked. Now it provides a more precise control of client traffic/sync traffic ratio on sync channel during nodes synchronization. For existing devices this parameter will be re-set to 50% after software update.

Tape Redirector

SPTD mode added for accessing physical tape device on the server. This mode bypasses user applications on the server. It suits the cases when the user software locks access to physical tapes for Tape Redirector module or breaks operations of remote clients that use iSCSI connection to physical tape device.
SPTD driver installation is required for this mode. SPTD mode is used automatically as soon as the driver is installed.

LSFS

Long device initialization issue has been fixed. Initialization of LSFS device with data after service start could take too much time, even after clean shutdown. Now the initialization time for device with data is significantly reduced.
Experimental version of LSFS module included. This version contains a lot of performance optimizations and the capability to store metadata on Flash storage instead of RAM

Virtual Tape

Added integration with Windows Deduplication feature for volumes where Virtual Tape Files reside.
Added option for storing Virtual Tape files on custom path.

Known Issues

When using StarWind with Synchrnous Replication feature inside of Virtual Machine on VMWare ESXi server, please stop StarWind service manually before pausing or taking snapshot of virtual machine where StarWind service runs.
Pausing the VM with running service under load may lead to split brain issues in devices with synchronous replication and data corruption.


Release notes V8 build 9996 (25 August 2016)

Asynchronous Replication

Fixed issue with the Management Console crashing when adding asynchronous replica to the existing device with synchronous replication, which has an old (V6) configuration format.
Old format assumes direct reference to .img file from HA device. Starting from V8 version HA devices use simple device as a storage.


Release Notes V8 build 9980 (23 August 2016)

Virtual Tape Library

Fixed issue with incorrect reporting of tapes in library for backup software. Device identifiers generator changed to use unique identifiers for new devices. Before the fix, IDs could be reused and client software could obtain incorrect information about VTL tapes in some cases.

Flat storage

Device properties reporting changed to allow Bitlocker encryption.


Release Notes V8 build 9781 (30 June 2016)

Cache

Fixed issue with service crash while deleting a device with large cache under load. Service could crash in some cases, when a device had RAM cache in write-back mode and was loaded with writes, while the size of cache was relatively large, several gigabytes, and user deletes device using Management Console.

Flash Cache

Experimental version of cache module added to this build. This module contains optimizations, which significantly improve performance of L2 cache.
Service can be switched to experimental cache module with Management Console: select the server, switch to Configuration page, Experimental Features section.
Switching to experimental cache module affects operation of L1 cache as well.
Experimental features are not recommended for use in production environment.

Flat storage

Size limitation of 16TB for single device is removed. Now the size of flat storage device is limited by size of file on the underlying filesystem.

Core

Fixed issue with processing of network configuration changes on server.
In case when network interface is being turned on and off periodically during certain period, service could crash on processing changed list of network interfaces for iSCSI server.

Web access to Management Console

Please check our web site for more information on StarWind Web Console feature.


Release Notes V8 build 9611 (2 June 2016)

Asynchronous replication

Asynchronous replication implementation reworked to use scheduled snapshots.
Source node creates snapshots by schedule and these snapshots are replicated to remote node. Local snapshots that have been replicated are deleted.
The number of snapshots to be kept on synchronous and asynchronous node are configured at the stage of replica creation.
Source device can be Flat storage or LSFS storage, replica is always created as LSFS device.
In case the source node is Flat device, additional disk space for write log is required.
Minimum network speed required for asynchronous replication operation is 100 MBps.

Notes

2+1 scenario: In case the source node for asynchronous replication is lost completely and second synchronous node remains functional , it is necessary to execute set up for asynchronous replica again using remaining synchronous node as replication source.
Upgrade from previous version: existing asynchronous replicas must be re-created after update to this version to ensure proper operation.

Synchronous replication

Added cases when fast synchronization is used instead of full synchronization. Fix for processing of underlying storage failures for case when Flat devices are used with synchronous replication. In some cases client could get disk error when storage failed on one node and is healthy on other node.

Cache

Fix for issues that occurred during creation and removing of devices with large cache size (several gigabytes and larger). In case there were other devices under load on the same server, those devices could process client requests with delays during creation or removing of device with large cache.

Tape Redirector

CDB length now set to 12 for vendor-defined SCSI opcodes (0xC0-0xDF) - to fix issues with SPTI access to some tape drives.


Release Notes V8 build 9052 (19 May 2016)

Core

Fixed error in cache module: cache memory could be paged to swap file in cases when cache size is more than 4GB.


Release Notes V8 build 8730 (11 November 2015)

Updates in this build:

Management Console

Fixed error with processing Cluster configuration records. If previous version was installed having configured Clusters in Management Console and installation was then updated to build 8716, Management Console crashed in a short time after start.

Release Notes V8 build 8716 (6 November 2015)

Updates in this build:

Management Console

Files of devices are stored in separate folders to simplify management.
L2 cache settings for synchronous replica can now be set up on replica creation without issues.
Health Status section added to device properties with synchronous replication to simplify sync channel status monitoring.


Cluster section has been removed. It will be part of new management functionality, which is being developed now.
Minor changes.

Synchronous replication

Solved issue where RAID rebuild on the currently synchronizing node affected performance of other nodes.

If active node is being synchronized and its partner detects that is has limited performance, it stops synchronization and retries attempt after a certain timeout.
Optimizations for operation with LSFS devices.

Cache

Cache size limit of 600GB removed.

LSFS

Fixed error processing when mounting device from existing files.
Performance optimizations conducted.

Fixed issues of snapshots management, which in some cases (example: removing device with mounted snapshot) could lead to service crash.

Processing of errors of underlying disk fixed, errors are now translated to client correctly.
Added command that starts forced defragmentation of device.

Important: It is recommended to move data from existing LSFS devices, created on previous versions of the software, to newly created devices after update due to changes in internal data structures.

Flat storage

Performance fixed. Disk access mode changed to use internal disk cache more efficiently. On some SSD drives and some RAID controllers, this error caused lower write speed with small data packets.

Core

Service now requires 64-bit OS for installation. Management Console, VSS provider and StarWindX can be installed to 32-bit OS or to 64-bit OS.

Known issues

For case of replicated devices, VAAI command may introduce overhead due to additional data synchronizations between replicas.
It is recommended to remove CHAP authentication settings before converting single-node device to replicated device. CHAP settings can be applied again once replica is added.


Release Notes V8 build 8198 (11 June 2015)

Fixes and updates in this build:

Core

Fix for error with ACL management: when access rights are edited in management console and service is rebooted after that, some access rights disappear.

Management Console

Fix for error with empty Browe window. It relates to old installations, which have been updated from version 5.8 to v8.
In some dialogs, where management console shows browse window to open the file or select file name for the new file, open dialog is empty and does not contain files and directories of the server.
It was caused by changes in default access rights for file browser in v5.8.

Virtual Tape Library

Added new virtual device type, MSL8096 (4 tape drives and 96 slots)
Added option to split virtual tape image file to parts for more effective file management.
Native Microsoft Windows deduplication works more effectively with files of size less than 1TB (https://msdn.microsoft.com/en-us/library/hh769303%28v=vs.85%29.aspx).
Added option to set custom size for virtual tape image.
VTL management functionality added to StarWindX (PowerShell module and COM-object).

Replication

3-node replication on LSFS: fix for issue when nodes do not synchronize correctly after fail in some cases.

Licensing

Error reporting for license registration procedure has been extended to make it more informative.

Cache

L2 cache: stability issues are fixed, library works as expected now.

Release Notes V8 build 7929 (17 April 2015)

Fixes and updates in this build:

Virtual Tape Library

Fix for error, which led to service crash on tape unload in some conditions.

LSFS

Deduplication option enabled. There was an issue with referencing deduplicated blocks, now it is fixed.

RAM Cache

Several optimizations to make cache working with optimal performance in cases of cache miss. Implemented asynchronous processing of SYNCHRONIZE_CACHE command, now it doesn't block command queue during execution.

Known Issues (fixes are planned for the next update):
3-node replication on LSFS: in some cases nodes does not synchronize correctly after fail.
L2 cache: stability issues found in current implementation.
Initial synchronization for Asynchronous replica for LSFS device can’t be executed after creation.

Release Notes V8 build 7774 (26 February 2015)

Fixes and updates in this build:

Core

Added support for new license types (check the Web site for more information).

Fix for processing device records that reference missing header files. Now they are shown in device list as non-active. If device files are removed from disk manually while service is not running, they was not showing in device list after the service is started. This could lead to error, when configuration file contains several records with the same device name (for example, several "ImageFile1" records).

Replication Manager

Fix for issue with switch partner operation. If device has been created in V6 or older version of StarWind, and partner is changed with V8 replication manager, new partner received wrong Vendor ID and Product ID.

It was needed to change these values in new partner's header files manually. Now the issue is fixed.

LSFS

Performance optimizations.

Defragmentation issues fixed. In some cases, defragmentation process could stop, and device files take more disk space then required. Currently during the normal operation, device files can take disk space up to 3.5 times more compared to nominal device size. Snapshots increase disk space requirements by unique data size in snapshot.

Deduplication option temporarily disabled for new devices.

Release Notes V8 build 7509 (29 December 2014)

Fixes and updates in this build:

Core

Fixed processing for UNMAP/TRIM commands. Virtual device was failing in cases when underlying storage doesn't support UNMAP/TRIM commands, but client use these commands to manage storage.
In the same situation, Windows system that is connected to Target with iSCSI Initiator was reporting to system log event with ID 7: "The device, , has a bad block".

Release Notes V8 build 7471 (11 December 2014)

Fixes and updates in this build:

Fixed processing for UNMAP/TRIM commands. Virtual device was failing in cases when underlying storage doesn't support UNMAP/TRIM commands, but client use these commands to manage storage.

Release Notes V8 build 7354 (5 November 2014)

Fixes and updates in this build:

Core

Fix for issue with service stopping. In some cases service could hang up when stopped under high load. In case of update from build 7145 it's recommended to disconnect clients before update to prevent old version from hanging up. Fix for processing VAAI XCOPY command. When used with VMWare vSphere, virtual device could fail during migration of virtual machines. Performance optimizations for SSD-based flat devices.

Caching

Fix for working with extremely small cache size.

Management console

Fix for error on managing target groups.

Replication

Fix for saving changes in list of network interfaces.

Virtual Tape Library

Fixes for reporting device properties to client software. Now different serial IDs are used for tape drive and changer devices.

Release Notes V8 build 7145

Fixes and updates in this build:

- Performance fixes for ImageFile (including usage with SSDs).

- VTL now works with MS DPM. It still can’t make backup that use several tapes with DPM.
It is not major issue, as DPM has the same problem with other tape libraries ( for example, Falconstore VTL).
Also BackupExec and ArcServe write multi-tape backups to StarWind VTL without any issues.

- Replication manager: fixes for networks validation on adding third node. In some cases this error led to error messages on adding third node to 2-node HA devices.

- Replication manager: ability to select network interfaces that reside in different subnetworks.

- Physical tape autoloader export: in some cases tape drive selection list was empty on the second step of Autoloader exporting wizard. Now filtering can be turned off in this list to make possible selection of tape drive.

- Fix for error on updating existing 6.0 HA devices that use CHAP authentication.

- Fixes in GUI.

Release Notes V8 build 6884

Fixes and updates in this build:

Licensing:

- Free license now have no limitations for single-node devices, as it was in V6.

Management interface for Virtual Tape devices:

Fixed bug with display of Virtual tape devices in Management Console.
They could be created but was not displayed properly in case of Virtual SAN for Hyper-V license.

LSFS container:

- Fixed bug in LSFS plugin. Deduplication could stop processing duplicated blocks in some cases.

This had two consequences:
    1) slowdown of storage performance on writes
    2) deduplication ratio stays around 1-1.1 even when disk stores a lot of duplicated data.

- Fixes on LSFS container compatibility with beta-2 and beta-3 versions.
In some cases LSFS devices that are created with beta-2 and beta-3 versions of StarWind could not be mounted properly with release V8 version.

Core:

- Fixed issue with processing of stacked devices on license registration.
Stacked devices are devices which use one or more underlying simple devices. Example: HA on LSFS with L2 cache (ImageFile).
If service started without license and license is registered, these devices become functional only after restart of the service.

- Fix for error with processing information about physical format of existing physical devices on server.
Service could crash on start when some of existing storage devices report incorrect sector size with values more than 1 megabyte.







Request a Callback