Release 3.2.30 * Performance improvements * Upgrade to 6.6.66 FC driver * Build fixes for recent distros Release 3.2.29 * Fix performance regressions introduced in 3.2.28 * Improve read/write performance with 4K logical sector disks * Fix service start failures on Debian Release 3.2.28 * Switch to Linux 6.6 QLogic FC driver for all recent distributions * Performance improvments * Fix possible deadlock when an iSCSI connection is closed * Improve read/write performance with 4K logical sector disks * Upgrade to lz4-1.9.4 Release 3.2.27 * RHEL 9 and Debian 12 support Release 3.2.26 * Switch to kernel 5.15 driver Release 3.2.25 * Switch back to 4.9.x driver from 5.10 driver Release 3.2.24 * Fix memory leaks which caused an oom condition Release 3.2.23 * Fix out of memory conditions due to daemon increasing virtual memory * Updated FC driver * Fix service startup failures * Fix identification of VMware disks which now report a valid SCSI device identification Release 3.2.22 * Fix GUI issues due to regressions in 3.2.21 Release 3.2.21 * Hotfix - Fix decompression issues introduced in 3.2.20 * Fix infinite loop when trying to read from a socket with error * Performance improvements * Clustering fixes Release 3.2.20 * Switch to default authentication by daemon * Performance improvements * Mirroring bug fixes Release 3.2.19 * Fix dbrecover to use new DB connection pool * Fix UI layout for "Access Management" page * Fix UI to redirect to vrlogin instead of invalid vtlogin.cgi Release 3.2.18 * Update to recent distributions * Merge fixes from VTL * Add dedupelite and disable dedupe options Release 3.2.17 * Speed up loading of deduplication tables * Allow reload of quadstor.conf by /quadstor/bin/vdconfig -r * Fix service shutdown hang on Linux. Earlier the sock_shutdown() was not called correctly causing a socket thread to wait forever. * Allow both upper and lower case when specifying a WWPN to fcconfing Release 3.2.16 * Improve SHA computation performance (upgrade to openssl 1.1.1a) * SLES SP3/SP4 and SLES 15 now supported Release 3.2.15 * Reduce ddtable syncs further * Fix read cache size computations * Reduce memory allocation for systems with < 16GB of RAM Release 3.2.14 * Reduce ddtable syncs * Fix build breakage on Ubuntu 16.04 * Fix possible list corruptions in FC driver * Fix reported soft lockups during ddtable loads Release 3.2.13 * Add option to align deduplication tables sequentially on disk * Add back support for SLES 11 and 12 * Create deduplication tables sequentially on disk * Use copy_page instead of memcpy in vdisk local read/writes Release 3.2.12.1 * Fix mirroring start failures for new VDisks Release 3.2.12 * Improve write-performance * Fix synchronize cache cmd stalls * Show mirror resync status in 'qsync -l' * Reduce the number of metadata reads * Move COMPARE AND WRITE commands (ATS commands) to a separate processing queue. * Support for newer Linux kernels (4.9.x) Release 3.2.11 * Fix kernel panic due to memory double-free during qmirror Release 3.2.10 * Fix automount of VDisks * Fix RHEL 7 and Debian 8 uninstall failures Release 3.2.9 * Increase XCOPY segment length to 256MB from 2MB. ESXi 6.x onwards requires handling of bigger segments for XCOPY operations Release 3.2.8 * Fix kernel crash introduced in 3.2.6 Release 3.2.7 * Fix out of memory errors with large VDisks (2) Release 3.2.6 * Improve write performance * Fix out of memory errors with large VDisks Release 3.2.5 * Add systemd service unit for Debian 8 and RHEL 7 * Increase metadata sync and free under low memory conditions * Add certain commands (INQUIRY, TEST UNIT READY) etc to head of queue for processing Release 3.2.4 * Fix IO stalls due to agressive write caching Release 3.2.3 * Rollback write cache performance changes * Fix slow FUA writes and IO hangs Release 3.2.2 * Improve FUA write and SYNCHRONIZE CACHE performance * Improve diagnostics * Add dependency to iotop, sysstat and lsscsi Release 3.2.1 * Fix possible sleep under lock when freeing vdisk metadata. Regression introduced in 3.2.0 and can cause READ/WRITE io delays Release 3.2.0 * Write performance improvements * Reduce system load average Release 3.1.88 * Decrease the number of clone/mirror threads * Increase back ddtable load wait time Release 3.1.87 * Fix memory leak during a duplicate block lookup Release 3.1.86 * Fix hung tasks when inserting new hash entries (ddthread_run) * Fix scanning of LVM volumes with hypens in the LV name Release 3.1.85 * Fix kernel crash due to NULL pointer dereferencing/Array index overflow in ODX implementation Release 3.1.84 * Fix double free for SCSI RECEIVE COPY STATUS when command is received for a VDisk in slave role * qla2xxx (FC) driver fixes for task management functions Release 3.1.83 * Fix vdconfig regression. Effectively vdconfig only worked for lists of VDisks due to the regression Release 3.1.82 * (FC/qla2xxx) Fix system crash due to a NULL pointer dereference when handling task management functions * Allow disabling of initiator-mode in the FC driver * Fix possible CPU soft lockup during writes when available disk space is low Release 3.1.81 * Fix memory leaks and kernel warnings on a FC loop down/up sequence * List individual vdisk statistics with vdconfig command line (vdconfig -l -v * Small performance improvements Release 3.1.80 * Kernel 3.19.x support. Includes support for Fedora 21 * Merge qlogic driver to upstream 3.19.5 Release 3.1.79 * Handle disks with older format NAA ids Release 3.1.78 * Fix soft lockups when cloning (Regressions introduced in 3.1.76 and 77) * Small write performance improvements Release 3.1.77 * Allow bdconfig to fix disks whose serial number or product id have changed. * Optimization disk metadata caching logic * Install html files in the correct location for Ubuntu 14.x * Allow configuration based increase of deduplication tables beyond the limit set based on the existing RAM size Release 3.1.76 * Fix invalid comparision checks when determining if tables need sync Reduces the number of syncs. Release 3.1.75 * Fix deduplication table insert warnings * Fix split-brain regression introduced in 3.1.74 Release 3.1.74 * Versioning related fixes * A clean shutdown of a VDisk in slave mode, should not try to fence master * During shutdown, commands which complete with error should have SCSI STATUS BUSY set instead of a CHECK CONDITION so that the command can be retried on another path Release 3.1.73 * Change versioning from 3.0.x to 3.1.x * Fcconfig rule fixes where rule for a specify HBA applied to all vdisks * Fix for vdisks not being detected over FC after a service restart * Fix for write failures to a mirror vdisk in a slave role when slave is resyncing from master. Release 3.0.72 * Fix false invalid number of ddtable entries reported Release 3.0.71 * Reduce the number of times the free block thread scans for unalloced blocks * When block allocator metadata cache overflows try harder to free unused indexes. Fix for possible out-of-memory scenarios * Optimize block allocation giving priority for disks with higher free space Release 3.0.70 * Fix kernel panic during deduplication tables flush, shutdown and service startup * Fix assertion during scanning for deduplication blocks * Optimize deduplication metadata loads Release 3.0.69 * Deduplication tables read write optimizations Release 3.0.68 * Allow upper case for WWPN and GUID with fcconfig * Fix kernel crash during cloning Release 3.0.67 * Emulate FUA and write cache support * Fix for huge page leaks with recent (3.10+) kernels * Fix for inaccessible out after a QUADStor service restart when accessed over the FC/InfiniBand interface * Handle disks (LIO/ReadyNAS for eg) reporting a serial number length greater than 196 bytes * When peer has failed and then try to reconnect ensure that we pick up newly connected sockets. Earlier a mirroring connection failed due to stale sockets Release 3.0.64 * Show LUN numbers when listing VDisks * Fix for incorrect FC rule ALLOW checks * Present consistent LUN IDs across restarts/reboots * Enable 512e by default in the 'Add VDisk' HTML page * Report that LUN is not ready and initializing when startup isn't complete Release 3.0.63 * Fix possible inconsistencies between master and slave due to write bitmap optimizations. Issue occurs when master has write bitmap enabled, mirror configuration is deleted and then reconfigured. When mirror configuration is deleted, the write bitmap is no longer valid, however we refer the same bitmap on a sync to slave after the reconfiguration * When deleting a VDisk we need to wait for all references to the VDisk are first decremented Release 3.0.62 * When multiple vdisks have active IO pending and if the peer fails, some vdisk in 'Slave' role fail to switch over to 'Master' when fencing was successful Release 3.0.61 * Fix for ITF package install failures * Fix for resync failures when a VDisk has a SCSI reservation present. Release 3.0.60 * Instruct browser not to cache HTML responses * Use SSSE3 optimized SHA-256 for Linux systems if the instruction set is available. Upto 10% speed up in hash computation is possible. * Miscellaneous build fixes for Ubuntu, RHEL 5 * For iSCSI read data digest errors Release 3.0.59 * Add checks for a VDisk's previous mirroring role to resolve conflicts where both VDisks drop to slave role. The previous role helps determine which VDisk is elevated to master role Release 3.0.58 * Reduce network reads when reading from a slave vdisk during resync * Fix false write errors reported to a host when writing to a slave during resync * Fix possible deadlock scenarios during mirror error recovery Release 3.0.57 * Merge lz4 r109 compression code * Merge srpt code to upstream * srpt abort command fixes Release 3.0.56 * Wait for VDisk mirroring connects to peer before enabling iSCSI listen port. This ensures that ESXi reconnects on dead paths when the are available Release 3.0.55 * Reduce the number of reads when writing unaligned data * Fiber channel driver merge with upstream linux 3.12.3 * Removed 'Transfer Rate' statistics, they were incorrect Release 3.0.54 * Fix VAAI Full Copy regression, which no longer worked due to ODX commits Release 3.0.53 * Merge lz4 r108 compression code * Integrate offloaded data transfers (ODX) support Release 3.0.52 * Synchronous mirroring error handling fixes * OpenStack cinder volume driver and related changes Release 3.0.51 * Fixes which allow for using VAAI full copy/XCopy during svmotion also. Release 3.0.50 * Fixes for new install and build system * Handle warnings when cloning VDisks which were resized * Fix for iSCSI IncomingUser truncation on a restart * Additional VDisk statistics * On a restart wait for upto 12 minutes for ddtables to load * On a VDisk rename, update IQN if its the default * Avoid sending threshold alerts to fast. Pause for 5 seconds before sending next alert * In SCSI Control Mode Page, indicate that task sets are maintained per I_T nexus * Extended VDisk names, which increase the maximum length of VDisk names to 256 Release 3.0.49 * Async load of deduplication tables. This was removed sometime back due to performance issues. * Fixed possible data corruptions with VDisks greater than 16TB in size due to log entries read resulting in a truncated LBA. log entries are read on a prior system panic, unclean shutdown etc. * Added checks for unsupported physical disk sizes (> 128 TB) and unsupported sector sizes (> 4096) * Further improvements to dedupe tables sync time during shutdown Release 3.0.48 * Improve dedupe tables sync time during shutdown * Fixed missing tdisk reference decrement during log pages replay * Added missing dedupe statistics increment, due to which the dedupe ratios shown were lower than actual Release 3.0.47 * Support for configuring multipath disks * Fix for configuring physical disks with serial numbers > 32 bytes * Fix for physical disks which return vendor specific device identifiers * Allow modification of serial numbers through vdconfig Release 3.0.46 * Fixes for running on Linux 3.10.x kernel * Regression where node controller did not send the VDisk list to node clients is fixed * Regression where VDisks are listed as physical storage is fixed Release 3.0.45 * Fix for limiting the number of dedupe table entries created. In certain cases we were not limiting based on the available memory * Fiber channel fixes and merge from upstream driver Release 3.0.44 * Cloning across pools with compression enabled did not deduplicate correctly. This has been fixed * Add more checks for received data count in mirroring message headers * bdconfig did not respect compression, log or ha disk options. * Fixed qsync usage message. Removed cancel option as it does not apply Release 3.0.43 * Changes required to allow installing the VTL product on top of the virtualization product * Fix bad page access when verify is enabled and a VDisk is written to using the local block device interface * Handle mirroring failures when VDisk thresholds are hit * iSCSI configuration fixes Release 3.0.42 * After a mirror resync the slave VDisk data may be inconsistent from the master VDisk data. This has been fixed * A mirror resync now no longer requires scanning of the enter VDisk metadata to determine which blocks of VDisk data need to be resynced * Command line tools for disk, VDisk, and pool management * qmirror inititated mirroring can lead to a deadlock if the source VDisk has active IO. This has been fixed * Source VDisks in a qmirror operation can now also be VDisks with synchronous mirroring configured Release 3.0.41 * Fixes related to Persistent Reseveration All Registrants Type * Fix for defects reported by Coverity Scan * Add scanning of Fusion IO devices on Linux * Improve speed of addition and deletion of physical disks Release 3.0.40 * Fix regression introduced in 3.0.37 where rcache entries need to be zeroed on alloc. Caused for invalid page frees * Persistent reservation fixes where service action CLEAR did not clear existing registrations and PREMPT did not clear existing reservation Release 3.0.39 * Fix deadlock introduced in 3.0.37. * Fix defects reported by Coverity Scan Release 3.0.38 * Open source release * Remove licensing restrictions * Build with debug enabled by default Release 3.0.37 * Fix for a major issue where in trying to delete an offline VDisk deletes an active VDisk Dont allow deletion of an offline VDisk or a VDisk which is being deleted Release 3.0.36 * Added Infiniband (srpt) target driver * Added support for 16 Gb FC (RHEL/Centos 6.x, Debian 7, SLES SP2 only) * Added Debian 7 support * Added FreeBSD 9.1 and 8.4 support Release 3.0.35 * Fix possiblity of double frees * Fix possiblity of manipulating lists when still accessed by another thread/handler Release 3.0.34 * Hotfix for kernel panic when using 512e VDisks with ESXi, Xen etc * List mirror configuration and configuration files in diagnostics * Warn if mirror fence configuration was not found Release 3.0.33 * Fix read errors on FreeBSD when compression is enabled due to bogus pages * Fix slow read/write performance on FreeBSD when compression is enabled * Reduce number of bio allocations for md raid 5 devices (linux) This should improve write performance for md raid 5 * More sanity checks in the dedupe code * Improvements to dedupe ratio * Fix possible hung tasks during synchronous mirroring error recovery * Fix possible hung tasks during block allocation when storage space is less * Handle extra spaces in configuration files Release 3.0.32 * Handle cases where one node can compress a block while the other node cannot * Improve read performance for local block devices * Check only /etc/fstab for automounting VDisks on Linux * Fixed bug where FC interface may report 1 LUN less than asked for. * Remove FC rules for a VDisk from kernel module when its deleted Release 3.0.31 * Speed up deletion of large VDisks. Deletion was too slow on Linux * Reduce memory utilization during VDisk deletion and cloning Release 3.0.30 * Minor changes to format of strings in trap notifications * Display better mirror status information * Synchronous mirroring bugfixes Release 3.0.29 * qmirrorcheck add -t "manual" * Fix qmirrorcheck parser errors for -t "ignore" * Add /quadstor/bin/qsync program to manually switch VDisk roles * Fix possible infinite wait during recovery of a peer failure * Add 'Compression Hits' stats which can be used to determine the compression ratio Release 3.0.28 * Run service restart using 'at now + 1 minute'. This fixes the hang when restarting the service through the GUI * Fix possible failures running pgpatch due to write permission of /tmp/qstorpgdb.log Release 3.0.27 * Fix regression where in VDisks setup for synchronous mirroring have different serial numbers Release 3.0.26 * Allow restarting system from GUI * Allow restarting service from GUI * Allow VDisk iSCSI IQN modification from GUI * Fix for a small memory leak when loading disk metadata * Allow specifying destination Pool when cloning a VDisk * Check for conflicts for custom default pool name * Compact "System Managment" page using a multi column layout * Compact "Modify VDisk" page using a multi column layout * If a memory allocation failure occurs during a read, while the command terminates with errors, further reads/writes can indefintely block. This has been fixed * Send notifications as SNMP traps Release 3.0.25 * Improve diagnostics reports * Add diagnostics for node clients * Run mount on VDisk attach and umount on VDisk delete/shutdown http://www.quadstor.com/support/139-automounting-a-filesystem-created-on-a-vdisk.html * Fix kernel crash if a read command times out on a controller node * Fix infinite waits during metadata sync (shared storage HA) * Fix metadata warnings during HA takeover (shared storage HA) * Specifying a fence command is now mandatory for shared storage HA * Increase the number of dedupe worker threads * Improve startup performance * Fix connect failures during mirroring setup Release 3.0.24 * FC drivers updated * Fix issues in cloning, resyncing mirroring slaves and deletion of VDisks. Certain LBA calculations can mean that trailing blocks are not cloned from the source, resycned to a mirroring slave or freed during delete. http://www.quadstor.com/support/downloads/Release-3.0.24.txt * Provide packages for FreeBSD 8.3 in addition to 8.2 Release 3.0.23 * Handle deadlocks during writes for VDisks whose pool depends on the Default pool for logging. This bug is easily reproducible during large unmap/discard operations * Remove warnings during dedupe operations due to stale entries for disks which were previously unconfigured * Handle log replay failure if any of the log entries contain references to a disk which was previously unconfigured * Fix buffer overflows during mirroring setup Release 3.0.22 * Fix synchronous mirroring timeouts * Improve error handling on mirroring timeouts * Fix false out of space errors * Fix/allow addition of SATA disks on FreeBSD 9 Release 3.0.21 * Synchronous mirroring bug fixes Release 3.0.20 * Fix timeouts during synchronous mirroring * Speed up VDisk deletion * Fix deadlock after a VDisk size reduction Release 3.0.19 * Fix for bugs in synchronous mirroring switch over from master to slave and vice versa. * Fix for regression where resync write data when a slave node is back online does not happen * Fix for possible deadlock in mirroring, clustering code when a remote node exits * Fix slowness in read/write when quadstor starts and vdisks are accessible but dedupe tables are still being read from disk * Added udev rules to create /dev/quadstor/ links on linux Release 3.0.18 * Fix quadstor-client package dependencies * Run builtitf during post install of quadstor-itf package Release 3.0.17 * Return SPACE ALLOCATION FAILED... sense on out of space condition * Allow configuration of Soft Threshold * Send Thin Provisoning Soft Threshold alerts * Fix regressions in mirroring and clustering code leading to possible deadlocks * Allow reset of VDisk stats in UI * Allow configuration of disk unmap and write caching in UI * Fix kernel crash during metadata sync (shared storage HA) Release 3.0.16 * Allow resize and rename of VDisk from the GUI * Allow rename of Pool fro the GUI (non Default pools only) * Fix possible deadlocks in write code Release 3.0.15 * Handle case where group/user scdbuser already exists when installing for the very first time * Support for ZFS zvol (experimental). Refer System Requirements * Support and use LZ4 compression * Significant improvement in write peformance * Fix warnings during HA switchover Release 3.0.14 * Handle updates to RecvAddr in ndrecv.conf for VDisks with mirroring configured * Handle conditions where a VDisk is not sure of its mirror role * Fix possible deadlock when a Persistent Reserve OUT command is received simultaneously by both VDisk mirror peers Release 3.0.13 * Reduce memory usage by core data structures * Fixed client node crash when VDisk properties are modified * Improve dedupe ratio when using EXTENDED COPY (XCOPY) and qclone * Fixed clone target VDisk data integrity issues if the source VDisk has compressed blocks * Fixed possible race condition during shutdown causing infinite wait on kernel thread shutdown * Fixed bug in XCOPY implementation when source and dest of copy are different VDisks leads to data integrity issues. This was a regression introduced by 3.0.12 Release 3.0.12 * For linux local block devices, a rescan-scsi-bus.sh used to list many virtual disks than available. This is now fixed * xcopy fixed to consider storage pools * Issue UNMAP/TRIM to underlying physical storage when required (Linux only as of now) * Allow addition of disk partitions as physical storage * Allow addition of Linux LVM volumes as physical storage. * Fixed poor performance of the qla2xxx driver in initiator mode * iSCSI auto reconnect to targets after reboot/restart. Previously on a reboot/restart, a rescan HBA was required on ESX * Skip display of swap, PVS devices in the physical storage html page * Storage Pool Statistics * Background VDisk deletion. * Allow VDisk resize to be lesser than current size. * Remove target-mode for qla23xx and older cards. * FreeBSD 9.0 support * Support for node fencing for shared storage clustering HA * Many bugs fixed in HA and clustering code Release 3.0.11 * Remove beta tags for releases * Fix autoconnect to VDisks on quadstor/system restart * Added SLES 11 SP2 builds Release 3.0.10 * Hotfix for kernel panic when FC interface is connected Release 3.0.9 * Fix remote mdaemon segfault when a mirror operation is started * Add -g option for qmirror command * Handle SATA controller disks which do not respond to serial number INQUIRY * Rework qla2xxx driver for RHEL/CentOS 6.x to merge linux mainline target implementation. qla23xx/qla22xx support is dropped for CentOS 6.x driver. SLES and Debian to follow soon Release 3.0.8 * Fix kernel panic on VDisk delete when using FC interface * Fix HA takeover by controller from client * Fix ndconfig printing of invalid nodes on node client Release 3.0.7 * Fix temporary CPU hang of node master during HA switch when node controller has crashed * Ensure data consistency during HA takeover when controller has crashed * Fix warnings generated during metadata sync in a HA setup * Ensure pgsql database is really stopped before install/upgrade * Check and warn for iscsi_trgt kernel module on linux installations before starting quadstor Release 3.0.6 * Fix for invalid locking/unlocking which lead to issues during mirroring, deleting of a physical disk when a controller-master pair is setup Release 3.0.5 * Storage Pools feature * ndconfig (/quadstor/bin/ndconfig) utility for displaying clustering * configuration * New HTML UI layout using yui3 library. * When log recovery is in progress, a system crash or reboot should restart the recovery again on next startup Release 3.0.4 * VDisk resize feature * Do not display VDisk under physical storage after a rescan * Miscellaneous fixes Release 3.0.3 * Log disk addition/removal cleanups * Removed debugs and code cleanups Release 3.0.2 * HA and Clustering bugfix release. Many issues were fixed Release 3.0.1 * database recovery fixes * fix for data integrity issues when multiple physical disks are configured Release 3.0.0 * Clustering * High Availability * Cloning * Mirroring * Miscellaneous bugfixes and performance enhancements Release 2.0.45 * Checks for offline master disk, when adding a new disk * No longer display or consider disks of size <= 4GB * Deleting empty VDisks did not reclaim all the space back. This is fixed now Release 2.0.44 * Fix SVMotion failures which eventually lead to ESXi reporting datastores as corrupt Release 2.0.43 * FreeBSD build only. Fixed incorrect reads via local block device (NFS) * for VDisks with 512e enabled Release 2.0.42 * Fixed random crash while adding a new physical disk Release 2.0.41 * Added support for HP Smart Array disks controlled by cciss driver * NOTE: Once a disk controlled by cciss driver is added to quadstor * it will not be recognized if at a later point the disk is controlled * by the hpsa driver. This behavior may be fixed in later versions * Packages are not rebuilt for FreeBSD as it is not required Release 2.0.40 * Fixed incorrect reporting of capacity used for physical disks >= 2TB NOTE: Users are advised to do a fresh install rather than an upgrade if any of the physical disks used were of capacity >= 2TB * Fixed incorrect capacity displayed for physical disks with 4K sectors Release 2.0.39 * Release specific to Linux * Changed builditf installation to install qla2xxx driver and update initrd. Similarly on quadstor-itf uninstall original qla2xxx driver is restored and initrd is updated again Release 2.0.38 * FC driver fixes * Debian Squeeze (6.04) initial packages Release 2.0.37 * iSCSI read performance fixes and fix for iSCSI command time outs and I/O failures reported by VMware Release 2.0.36 * Primarily FC target mode fixes. FC targets did not work under certain conditions Release 2.0.35 * Fix rescan physical storage logic. Earlier under certain conditions the physical disks displayed contained duplicates, the number of duplicates increase on every rescan * Send addtional ctio for cmd overrun/underrun conditions. Fixes FC targets not being detected Release 2.0.34 * RHEL/CentOS 6 packages now available * Rescan of "Physical Storage" also lists created VDisks on Linux. This has been fixed * Fixed default runlevels for quadstor etc script on RHEL/CentOS 5/6 * Use netlink_kernel_release for newer linux kernels. Fixes crash during shutdown on RHEL 6 * Fixed vmware virtual disk detection on quadstor restart, physical storage rescan, system reboot Release 2.0.33 * Hotfix for kernel crash introduced by 2.0.32, due to delete VDisk changes Release 2.0.32 * VDisks are now reported as Non-SSD disks. On certain initiators/hosts (for example recent linux kernels) this can lead to a read performance degradation if the underlying physical storage isn't fully SSD. * Deletion of VDisks earlier had a high CPU utilization * Fixed installation failure in SLES packages