$Id: readme_win32_cli.txt,v 1.18 2007/11/05 11:09:50 sesam Exp $ SEP sesam Backup Client (SBC) V 3.2.X.Y on WIN32 - Windows XP, Windows 2000/2003 Installation Guide and Release Notes ==================================== 1. Installation Guidelines ========================== Basic Requirements: - Login with administration privileges. - Supported operating systems: Windows 2000, Service Pack 3, Home, Professional or Server Windows XP Windows Server 2003 Install SEP sesam Backup Client with the following steps: 1. Change into the CD directory :\win32 2. Extract the files with a call of win32_cli.3.2.X.Y.exe 3. The installshield will be started automatically. Please follow the instructions during installation. 2. Update Guidelines ==================== Please install kit for update with same steps as described under "1. Installation Guidelines". The installshield programm will recognize your installation and an update will be performed. 3. Release Notes ================ The following Release Notes are extracted from CVS (Concurrent Version System) and they are describing the source code changes. Kit 3.2.1.3 contains the following changes to version 3.0.1.12 *************************************************************** Setup and System ----------------------------- BUGFIX: Installation with pop-up message "Warning: Could not start SM_MAIN. Restart manually." was caused by failure during installation's restore test. sm_main 1.117 CHANGE: Support sqlite: Doesn't need a sm_db_main process CHANGE: Do only kill java*sm_rmigui.jar processes to avoid killing of other java applications sm_setup 1.288 BUGFIX: check return of 'gethostbyname' to avoid core during setup and no correct name resolver configured SEP sesam DB changed to SQLITE 3.3 ---------------------------------- BUGFIX: On Windows with multi processor a call of OpenAccess damisql could hang. Resolved by changing SEP sesam DB to SQLITE v3.3 SBC VERSION 2.4R22 ------------------ 2.4A22: Bugfix: Restore could break with BAD_ARCHIVE if 2 bytes were interpreted as alignment SBC_UC VERSION 3.0A8 (sbc for WIN32 Unicode) -------------------------------------------- 3.0A8: Increase stacksize bcs. last version 3.0A7 could lead to stack overflow 3.0A7: Bugfix: Restore could break with BAD_ARCHIVE if 2 bytes were interpreted as alignment 3.0A6: Bugfix: Padding remaining bytes if part of file is locked by a process was wrong in version 3.0A5 3.0A5: sbc_uc with bugfix avoid interpreting NTED in stream as NTED Header, Avoid overwritting of skipped files 3.0A4: Bugfix: In case of restore malloc( _MAX_PATH ) without free for vttrace() could lead to memory underrun and breaks process execution 3.0A3: Avoid message sbc-2046: Warning: Cannot get item security data for [...] for files which are opened for write. Avoid message sbc-2034: Warning: Item archive attribute resetting failed. Item: [...] for directories 3.0A2: With support of Encrypting File System EFS encrypted files SM_CTRL VERSION 1.6A6 --------------------- 1.6A6: Avoid command overflow in CTRLD SMS VERSION 3.2A7 ----------------- 3.3A2: STPD for Win32 Unicode: path in lis file for directory must end with / BUGFIX: (WIN32) On SLR QIC media sometimes a SetTapePosition() command requires a following GetTapePosition() otherwise the next command will fail, e.g. in case of restore after skipping FILEMARKS ReadFile() will fail after appr. 15 minutes timeout with error 1111 ERROR_BUS_RESET 3.3A1: WIN32 MTF stream with file name as wide char (Unicode) is now supported and converted to UTF-8 SEPsesam Kernel 3.2.1.3 ------------------------ NEW: Media duplication with Migration tasks NEW: Automatic saveset lis file (journal) compression with gzip CHANGE: Edition 'one' and 'Community' only with one real hardware drive, e.g. DISK_CHNG or tape device sm_backup 1.84 BUGFIX: Fix problem, if several backups ar started at the same time and the expected label, but another EOL-free one is in drive. Select and load of media is now completely protected by one semaphore. Before between select and load was race condition, so more then one backup tried to load the requested medium. sm_client 1.114 CHANGE: don't write error message in sesam protocol, if sm_client returns with error. sm_config_client 1.27 NEW: correct access check for VMS clients BUGFIX: set correct 'stpd_options' for cennection test to allow test with firewall between client and server sm_get_volume_lib 1.31 BUGFIX: in 'i_CATCH_STRANGE_MEDIA': sm_arch was called with wrong paramter list, so inititalization of SPARE medium failed BUGFIX: in 'iCHECK_SPARE_MEDIA': sm_arch was called with wrong paramter list, so inititalization of SPARE medium failed sm_init 1.154 BUGFIX: If sm_break sends interrupt during GET_MEDIUM() then a valid label could be overwritten by requested label sm_lib 1.514 CHANGE: GET_LABEL_VIA_SMS(): Return 3 if a non valid label was found, e.g. in case of Veritas tape label NEW: License for OX_GWUSER and SCALIX_USER NEW: compression of lis files: If 'compress_lis' is set in table defaults then lis files older then 'value' are compressed by gzip NEW: For DISK_CHANGE: call os specific cmd file 'sm_os_diskchange' (if exsits) to specify for example mount/unmount command on UNIX Parameters: $1 = device name $2 = unload|online Return: 0 = device is online NEW: Function 'i_UNCOMPRESS_FILE': check if file with '.gz' extention exists and uncompress it if necessary. Bugfix: in case of EOM and retry login with login type EOM CHANGE: i_LIC_INSERT_ITEM Compare licensed server name with sm.ini [SERVER] name=xxx ignoring the case BUGFIX: DELETE_OLD_LIS(): PostgreSQL: Old algo couldn't remove all results entries. Now all result_lbls referring deleted results will be removed first. sm_lis_db 1.82 BUGFIX: Avoiding ascii_filter (lis-file) leads to problems if CR is part of file name. Now call SYNTAX_CHECK (lis-file) and remove CR; If in sm.ini [PARAMS] lis_with_high_ascii=NO then remove high ascii chars too bugfix: CHECK_EXTRACT_DIR(): Change for SCALIX did not check for not connected directories which are needed for OX backup NEW: Support of compressed (gzip) lis files. If 'lis' file is compressed, call gunzip CHANGE: Support for SCALIX 'M' for mail and 'm' for mailbox CHANGE: Improved generation restore - much faster algorithm: Build generation restore lst file with postfixed line number to allow resort of sel file to order of lis file. (sm_restore.c 1.253.2.7) CHANGE: Call to ascii_filter( {lst file} ) removed to support unicode sm_newday 1.61 NEW: Call iDB_ACCESS('vacuum;',...) for sqlite DB CHANGE: If gv_db_type is postgres then call DB_ACCESS with 'VACUUM VERBOSE' to remove deleted rows sm_pur_status 1.46 NEW: compression of lis files: If 'compress_lis' is set in table defaults then lis files older then 'value' are compressed by gzip sm_restore 1.265 BUGFIX: Restore test during installation failed bcs. char szQue[] was too short and so submit command failed NEW: if 'clients.data_mover' is given in DB, execute 'sbc_smdr' on this host BUGFIX: include MTF volume descriptor block in all selective restores. Else relocation will fail, bcs. filename doesn't match relocation filter BUGFIX: during generation 'sm_restore' was called recursivly, if task name starts with 'y'. Now use '-g' switch for generation restore on command line. CHANGE: Correct update statement of parent restore task, so it does not stay active in database BUGFIX: Task type 'win32_unicode' with target '/ORIGINAL' lead to error 'Option -R requires an argument' bcs /ORIGINAL was not converted to '/'. Now target path is set to '/' if /ORIGINAL was appended to original path and backup type is 'unknown'. If target is empty string then we use '/'. NEW: recreate lis file for generation restore, bcs. in case of migrated savesets it does not exist. NEW: call last restore of generation restore and 'sbc_post_restore' with 'last' BUGFIX: Set target path to '/' if /ORIGINAL is appended to target path for task type 'NetWare', not only for task type 'Path' CHANGE: If restore_option is given then do not skip NetWare options. Build NetWare options and add restore option BUGFIX: In version 1.243 setting of gszWorkDir with gv_rw_work was removed but is needed for COPY_FILES_FOR_SEL() to tape_server CHANGE: OPEN_FILES_FOR_CREATE_SEL_FILE() Call i_UNCOMPRESS_FILE({lis-file},...) to uncompress if compressed CHANGE: RESORT_SEL() avoid multiple malloc() and free allocated pszTmp BUGFIX: In case of large generation lis files old algo was looping through sel and lis. This lead to N x M read line operations. Now we sort sel in order of lis file and we have only 2 x M readline operations ATTENTION: This algo needs new sm_lis_db sm_robot 1.129 CHANGE: Count errors instead of warnings and put error count message in the SEPsesam protocol change: Avoid writing: "E004-MEDIA [ 14467]: Adminstration of media-archive doesn't return a status." bcs this overwrites message from sm_arch CHANGE: Improved for DISK_HARD media: Do not delete new slots from loader_contents and set barcode to label sm_sbc_com 1.103 NEW: for DB2 replace '?' wildcard by '_' for SQL select sm_sbc_com_lib 1.18 BUGFIX: Fix problem, if serveral backups ar started at the same time and the expected label, but another EOL-free one is in drive. BUGFIX: iSUBMIT_EXT() After timeout the qm del_entry for external job was wrong. Now corrected: sm_qm que_del_entry {entry} sm_sepul_event 1.159 BUGFIX: If a backup was started then it's ev_type was set to 'C|F|D|I' and therefor a following backup task couldn't find an already done task with higher prio BUGFIX: Shared drives backup could fail with: "The requested volume %s does not correspond with the current volume in slot %s loader %s (%s)." if a concurrent backup was started with a given drive. Reason: CREATE_QUEUE_NAME(): If called for a given drive then '@tape_server' was not added to Queue name. sm_sms_backup 1.272 BUGFIX: If 'archiv_bit' was set to 'no' then backups for Exchange or SQL Server are started with -l copy -T {timestamp} instead of full /diff /incr now 'archiv_bit' is only checked for 'Path' Backup Bugfix: VMS Delimiter was not set correctly for Win32 bcs of missing break in switch case BUGFIX: In case of VMS and Linux OES NetWare backup COPY_REFORMAT_LIS() must be called with R->szBckSts set to "V" or "N" to process lis in a specific manner BUGFIX: If key='with_archive_bit', value='no' is set then WIN32 backup level FULL and DIFF is done without archive bit resetting. In old version -T switch was enclosed in too much " sm_sms_watch 1.114 BUGFIX: wait longer for trailer ready, bcs. sometime tapedrives need more time CHANGE: Exit sm_sms_watch if queue state is 'pending' after backup has finished bcs. in case of shared drive the state may be 'pending' instead of 'idle' sm_hal_wnt 1.211 bugfix: start_exe() has to return exit state of executed program, old version always returned 1 sm_kill 1.8 CHANGE: If called with a '*' in pattern then match with CLI, e.g. 'java*sm_rmigui.jar'; Read SM_KILL in debug.ini to set trace automatically; BUGFIX: Do not kill if pattern matches command line arguments only if pattern matches process sm_qm_main 1.58 NEW: Check preferred resources, e.g. drive number to get highest job for resource queue. Old algorithm blocked execution of jobs for a free resource if a job with a higher prio was waiting for another resource. NEW: If resource (drive) is blocked or stopped while queue is running then jobs with same preferred resource are changed to any '*'resource only if requested server is '*' or identical to resource's server. This will allow the Queue Manager to start pending jobs in other drives if requested drives are blocked or stopped. This feature is not used in current SEPsesam but may be used in future releases. Example: Stop drive 3 by explicitly calling sm_qm block with resource 3: #> sm_qm block Bandlaufwerke/3 After this waiting jobs for drive 3 will be changed to any drive. With stopping or blocking of the whole queue no changes were made to pending jobs, e.g. stop Bandlaufwerke: #> sm_qm stop Bandlaufwerke CHANGE: que_init now allows changing of counter for already defined resource queues BUGFIX: In case of shared drives the share and state of an existing drive queue (resource queue) must be changed by further calls of sm_qm init {queue}[/[{share}_]{drive}[@{server}] sm_sepuler 1.25 CHANGE: GET_TERMS: If DB_ACCESS() returns error then return 0 and do a retry sm_sepuler_cal 1.23 change: sqlite only accepts unix style directory separator / so use {gv_rw_work}cal_insert.tmp SEPsesam GUI 3.0 Build 2 R1.184 ------------------------------- - Adapted for SQLITE - Support for SCALIX - Support for Media Duplication - > Migration Task - Extended Calendar Sheet xxx.java CHANGE: For SQLITE Avoid method .wasNull() in many java sources ArchivDialog.java 1.18 new: memorize users settings for small slots view and use barcode view (references #3131) CalendarSheet.java 1.36 Support for all EventTypes with colors for state: added migration, task_groups, sm_startup CliBroDialog.java 1.63 change: no input dialog for 'system_state'; NEW: Support for browsing Scalix NEW: Client browsing with unicode support. Frame.java 1.72 NEW: if GUI client is started with '-M 0' option, master sesam function is switched off (regardless of entry 'mastersesam' in defaults) Especially interesting for remote access over SSH. LoaderActionDialog.java 1.15 CHANGE: Support large tape labels LoaderDialog.java 1.22 bugfix: if no loader was defined it was impossble to add a new loader ScheduleDialog.java 1.84 BUGFIX: Correct handling of interface by changing media_pool to a different tapeserver Release Notes for Version 3.2.1.14 ================================== SESAM DB sesam_db.sql 1.110 CHANGE: [backup_]|[restore_]options CHAR(128) NEW: table 'preferences' for saving user GUI position Kernel Library sm_lib 1.522 CHANGE: GET_LOCAL_TCPIP_ADRESS() returns a string with all local IP addresses to allow adding of ethernet cards without disturbing current SEP license bugfix: ANALYSE_RESTORE_LOG() If ( ilProcessedItems == 0 && ilNotProcessedItems == 0 ) then set state to 'X' and message to XX_SMSREST_NO_ITEMS_PROCESSED despite of 'WARNING' state Kernel Modules sm_client 1.117 BUGFIX: In case of dir for NetWare build sbc_smdr command without TSA name set to node name bcs. GroupWise has problem with lowercase TSA name. Now sbc_smdr inquires TSA name and uses first given TSA name sm_lis_db 1.86 bugfix: After call to squeeze() write buffer with strlen(buffer) bcs. squeeze might have removed some '/' CHANGE: In SYNTAX_CHECK() multiple '/' are squeezed otherwise subdirectories are not shown in GUI Restore Wizard, e.g. /var/work//tmp/scalix was only browsed til /var/work sm_pur_status 1.51 BUGFIX: POSTGRES delete sesam db backup files in 'gv_rw_db:/backup' and not from 'gv_rw_work:/backup' sm_restore 1.274 CHANGE: Support for AES encryption crypt_flag 'a' BUGFIX: Saveset migration: Do not change task type, if no 'restore_cmd' found in database. Else sbc is called with interpreter 'cpio' in case of an 'Oracle' backup. BUGFIX: Wait 5 seconds after submit of 'sm_sms_watch', because read from DISK device can already go into EOM, before 'sm_sms_watch' is really ready. sm_sbc_com_lib.c 1.20 CHANGE: retry check for 'sm_sbc_com_ext' job, bcs. QM sometimes not answering correctly sm_sms_backup.c 1.274 CHANGE: Check if sbc at client side supports AES encryption if AES encryption is requested: UNIX since sbc 2.0 sm_qm_main 1.62 BUGFIX: If max Entry Number was reached old algo was only resetting Entry Number to 1 but this could be wrong for still existing numbers e.q. 1 sm_postgres, 5 sm_ctrld_main, 6 sm_sepuler Now next entry number is checked and increased sm_setup 1.313 BUGFIX: Output of log file information in version 1.311 lead to problems with get_policy so User administration with GUI was impossible GUI V3.0 Build 5 R 1.243.4.5 CHANGE: Display RMI mode (classic or cajo) in about dialog BUGFIX: changes of end-of-lifetime were not always saved in db in the MediaPoolDialog bugfix: Dialog 'User rights' does not work if gui_request is sending sm_setup get_policy and this returns some trace code before the proper values. NEW: use protocol status as newday protocol in ResultsDialog BUGFIX: when a command event was shifted to another execution time the other events were not moved accordingly -- see #4068 NEW: SCALIX Backup Source: Warning if vgpath=none and no online backup will be possible. BUGFIX: command events where not deletable via right mouse click BUGFIX: correct path of netware path in case of client os is LINUX BUGFIX: path of exclude list was not correct in case of 'all' backups; pre and postfix of linux exclude lists where missing; BUGFIX: in Client Browser: The exclude list of a path backup was not built properly. Release Notes for Version 3.2.1.16 ================================== SESAM DB SQLite ---------------------------- CHANGE: Use CHAR instead of VARCHAR to build sesam_db.sql for SQLite bcs. VARCHAR is CHAR with variable length in SQLite Libraries ---------------------------- Now with libeay32.dll crypto library GUI V3.0 Build 7 R 1.243.4.7 ---------------------------- LoaderContents.java CHANGE: layout changes to make the loader content resizable; CHANGE: changed the internal refesh interval to 10 sec. CHANGE: avoid flickering of loader content NEW: refresh interval can be external set with key "loader_refresh_interval" in table defaults; ClientDialog.java BUGFIX: tasks where not saveable if data_mover is empty Globals.java BUGFIX: don't try a connection if connect=off; donīt try to terminate a non existing connection RemoteServerDialog.java BUGFIX: don't try a connection if connect=off; donīt try to terminate a non existing connection ScheduleDialog.java BUGFIX: in some cases the schedule dialog did not close properly. BUGFIX: Migration event: the source pool and the source drive have been selectable in the comboboxes SingleRemoteServerDialog.java BUGFIX: The remote server's client name was filtered twice -- removed the oblsolete standardStringFilter CliBroDialog.java BUGFIX: Handling of '/' if Pattern matches with "NetWare|Linux [Cluster] File System" MigrationByStatus.java BUGFIX: Subrestores of a generation restore must not be visible in Job State Kernel Modules ---------------------------- cm_lang.h BUGFIX: Misleading message for XX_LICENSE_ERROR bcs. in all languages except German same text like XX_LICENSE_ERROR_FILE_OPEN was printed: "W007-LICENCE Missing the licence-file - please contact support@sep.de" Now we use: "W014-LICENCE Wrong encryption: " cm_hal_pipe.c BUGFIX: iCreateAndOpenNamedPipe() Windows 2003 Server with SP2 requires to create named pipe with security attributes otherwise communication with pipe will hang, e.g. hanging sm_qm.exe cm_hal_wnt.c BUGFIX: i_setenv_sesam() was buggy since version 1.54 now PATH is set correctly again; NEW: Set OPENRDA_INI only if gv_db_type == "OpenAccess" cm_db_access.c BUGFIX: DB_SEL_GET: If argument is 'd' then we have to break for loop to avoid setting pointer to next element in already deleted list - possible segmentation violation sm_copy.c BUGFIX: Selection of saveset for migration take account of backup level again sm_ctrlc_lib.c BUGFIX: Avoid reuse of global file handles: FILE *fd_stdin, *fd_stdout, *fd_stderr, *fd_oldout, *fd_olderr; Otherwise output of an asynchron created thread may be blocked by a following call of ctrlc_lib_call(). Problem appeared in sm_sms_watch during handling of EOM situation on a Remote Device. sm_db.c BUGFIX: If called for export then write possible error message with -- error: {message} bcs. old style leads to problems in export file, e.g. message "STATUS=ERROR MSG=no such table: mediapool_locations" was the reason that task 'SESAM_BACKUP' was not inserted into sesam_db. CHANGE: Creating of extra smdb-{table}.exp files during export removed. sm_info.c NEW: if argv[1] is a number then print strerror({number}) sm_restore.c BUGFIX: copy segment file for restore also in case of saveset migration, to avoid problems, if tape trailer doesn't contain correct next label (EOM bcs. of IO error for example sm_robot.c NEW: Mediapools on disk can use more then one mount point (or directory) sm_sbc_com.c CHANGE: CBMR: iDB2_SEARCH_SAVESETS() return saveset info together with start_tim=%s sm_sbc_com_lib.c BUGFIX: correct return value, if normal backup job is started by sm_sepul_event sm_sepul_event.c BUGFIX: write backup interface into results during first insert, so we can use it for restarting failed backups with correct interface sm_sho.c CHANGE: In case of NOT/PRE/POST: If task is given instead of saveset then max(saveset) is used. sm_sms_watch.c BUGFIX: On Windows thread for fetching performance info from RDS must be stopped after EOM handling otherwise old output (e.g. 'EOM') could be read again sm_qm.c CHANGE: Use cm_hal_pipe.c standard ...NamedPipe(.) functions bcs. handling of timeout is not required sm_qm_main.c BUGFIX: In case of shared drives a job could start on wrong drive if drive was not given. Nw we check drive and server. BUGFIX: sm_qm del_entry went into timeout, if signaled program couldn't be terminated by singnal 2 (SIGINT). Now enable 'alarm' handler after 5 seconds again to kill process group by signal -9 CHANGE: Increased performance with IPC_VERSION with named Pipe communication BUGFIX: Queue synchronisation could fail, e.g. for shared drives. Now iSynchronCount is counted for every Queue. ctrlc.c >> sm_ctrlc CHANGE: Option -f (fake) added to allow return 0 even if OOB signal was not received ctrld.c >> sm_ctrld NEW: if 'auth' mode is switched off ('-a' switch set in sm_ctrld_main commandline), then do not exit, if gethostbyaddr() fails. So we can use 'ctrl' without any name resolution, but there is no authentication at all BUGFIX: Avoid endless loop on ctrld site if scc dies unexpectly. The loop could lead to a 100% CPU usage on Sesam client side. If select signals socket activity, but read returns '0' bytes, then other side has closed connection. sm_sepuler_cal.c BUGFIX: Last change for SQLITE makes export unusable with POSTGRES. As a result calendar sheet did not show future entries in case of POSTGRES. slu.c CHANGE: iLinuxScsiInquiry() Check for cBuf[0]==8 || (cBuf[0]&0x1f)==0x1f bcs. StorageTek may return 0x1f if LUN != 0 BUGFIX: If loader has a LUN <> 0 then in CDB for Inquiry LUN must be set apropriate (implemented for Linux) sm_main.c BUGFIX: sm_main start stop failed since revision 1.118 (v3.2.1.3) if SEPsesam client is installed in directory with blanks, e.g. \Program Files\SEPsesam In function iGetPid(): iPIPE_OPEN() does need command enclosed in " to support blanks in command path to get output of C:\Program Files\SEPsesam\bin\sesam\sm_kill list sm_setup.c BUGFIX: In case of _gui_ kit a client installation has to fetch version from skel/{os}_gui_.{main}.{minor} new: Support for very old SuSE 6.3 linux SOB ORACLE RMAN Backup Version 1.21 ---------------------------- libobk.c BUGFIX: Check if closesocket(...) fails with EINTR and if then retry sm_oracle_client.bat BUGFIX: Trace file given with SOB_TRACE=%trace% CHANGE: Allow rman backup with 'nocatalog'. Avoid checking of Target and rcvcat connect string to allow system user authentification stpc_func.c BUGFIX: stpd_port(): Bind LISTENER SOCKET to same IP address and not to 0.0.0.0 otherwise stpd may break the save operation SBCMSQL MS SQLServer Version 2.50.0.27 ---------------------------- CHANGE: Windows Server 2003 with SP2 requires setting of SECURITY_ATTRIBUTES to allow opening a named pipe Release Notes for Version 3.2.1.18 ================================== Major Changes ------------- BUGFIX: Avoid deleting whole matching directory tree, e.g. "C:\Programme" if matching with first part of SEPsesam directory, e.g. "C:\Programme (x86)" Installshield (setup.exe Version 343) BUGFIX: In case of single tape drive if backup detects wrong media then an init was started and there was no chance to insert already prepared media. Now wrong media will be ejected and next media will be checked. BUGFIX: Alive Test could fail or hang bcs. sbc -h to print usage could lead to a core dump on some systems. BUGFIX: Scheduler had problem calculating with DayLightSaving time and execution at 23:00 or 00:00 o'clock. BUGFIX: After Reboot creation of main queue qu_all could be aborted so main programs could not start in qu_all and remain offline BUGFIX: Init for drive group could fail bcs. SQLITE access for 'group by' was not handled properly NEW: Support for MS SQL Server 2005 Syntax for source changed to [/{SQL-Server}/{instance}/]{database} GUI V3.0 Build 8 R 1.243.4.11 with - Display of drive numbers in Topology - Migration task shows alert message if target drive is not set - Client Browser fpor NetWare and Exclude revised - Scalix support of different orders of target elements (mailnode=...:instance=...) - The calculation of monthly executions was wrong if locale was set to en; weekly calculation was not executed if a day was deselected; Austrian week started incorrectly with sunday; NEW: In table defaults a default interface for a drive group may be inserted: key='default_interface_', user_name='sesam', value='' e.g. key='default_interface_Tape_Drives', user_name='sesam', value='backup_2' NEW: In table defaults a default for the joblimit of au_all may be set: key='qu_all_joblimit', user_name='sesam', value='' e.g. key='qu_all_joblimit', user_name='sesam', value='40' The default value is 20. NEW: Special handling for device type VXA with variable blocksize CHANGE: Allow init in case of EOM with empty media BUGFIX: Correct priority check for saveset duplication BUGFIX: If a media from SPARE pool or an empty / unknown media was successfully initialised then the mount was not done so a second init could not get current label and tried to initialise for pool 'NULL'. Same bug was corrected in version 1.156 but due to changes in version 1.159 bug reappeared for SPARE or empty/unknown media. BUGFIX: Now xbsa supports calling program sbc_com_interface when path contains blancs e.g. C:\Program Files\SEPsesam\bin\sesam\sm_sbc_com_interface.exe CHANGE: sm_recover with higher performance due to sequential read of savesets from media SBC VERSION 2.4A26 BUGFIX: In case of restore with mode overwrite files will be opened even if SHARING_VIOLATION appears BUGFIX: sbc -h to print usage could lead to a core dump on some systems SBC_UC VERSION 3.0A21 (sbc for WIN32 Unicode) BUGFIX: In case of restore with mode overwrite files will be opened even if SHARING_VIOLATION appears NEW: Support for MS SQL Server 2005 Syntax for source changed to [/{SQL-Server}/{instance}/]{database} GUI V3.0 Build 8 R 1.243.4.11 with - Display of drive numbers in Topology - Migration task shows alert message if target drive is not set - Client Browser fpor NetWare and Exclude revised - Scalix support of different orders of target elements (mailnode=...:instance=...) - The calculation of monthly executions was wrong if locale was set to en; weekly calculation was not executed if a day was deselected; Austrian week started incorrectly with sunday; NEW: In table defaults a default interface for a drive group may be inserted: key='default_interface_', user_name='sesam', value='' e.g. key='default_interface_Tape_Drives', user_name='sesam', value='backup_2' NEW: In table defaults a default for the joblimit of au_all may be set: key='qu_all_joblimit', user_name='sesam', value='' e.g. key='qu_all_joblimit', user_name='sesam', value='40' The default value is 20. NEW: Special handling for device type VXA with variable blocksize CHANGE: Allow init in case of EOM with empty media BUGFIX: It was possible to insert a media type containing a space BUGFIX: A media type "50 GB" causes an error in sm_inventory, because of wrong parsing. Minor Changes ------------- BUGFIX: Support display of block size for saveset and media bigger then 2 TB CHANGE: Protocol messages for Migration now with correct indention.