Changes 4.4.2

From SEPsesam
Jump to: navigation, search

Major fixes and changes v4.4.2.88

Currently only released for Windows x86!

For a complete list of GIT change reports, see changes.

BSR Windows Pro

  • Windows BSR Pro for operating system architecture x86 (32 Bit) now supported.

Installation

  • BUGFIX: (#19952) SEP sesam GUI start requires 'Run as Administrator'. Problem: the java path which is needed for the GUI to start is adjusted using the sm.ini which can only be updated as administrator. Solution: Only obtain the correct java path without writing it to the sm.ini. Start the GUI anyway.
  • WORKAROUND: Do not use 'systemd' on Debian 8, because build script aren't adopted at the moment

Kernel

  • bugfix: Read SMS performance file with 'binary mode' to avoid 'E001-WATCH Received performance data with invalid format' errors, because file include CR/LF after 2048 bytes

Restore

  • BUGFIX: (#18178) File view for Dovecot doesn't work any more. Problem: 'IMAP:' entries are displayed as directories. Solution: Build entries for mailbox folders with type 'm'

SBC

  • BUGFIX: (#19827) VSS Backup ends with 'HardlinkCleanProcessedGroup: Hardlink ... not found'. Problem: Last solution failed to remove already processed hardlink - with changed major:minor - from list. Solution: If major:minor changed then remove hardlink via name from list
  • BUGFIX: (#19548) Warning during system state restore on UEFI systems: Cannot create item ... . Problem: Path 'GLOBALROOT\Device\HarddiskVolume2\EFI' refers to a volume not a real path. Solution: Skip volume part of given path

Major fixes and changes v4.4.2.87

For a complete list of GIT change reports, see changes.

Operating Systems

  • Support for Windows 10 32 Bit (x86)

GUI

  • BUGFIX: (#19648) NPE during saving of drive properties. Problem: hwLoader with invalid name will be inserted into DB. Solution: remote pattern of hwloaders.name and allow all
  • BUGFIX: (#19508) vCenter license is for Datacenter repeating vCenter. Problem: vCenter will be written again for each datacenter. Solution: clear vCenter buffer after write data to output
  • change: (#15908) set correct output format for list vm license; prevent log of datacenter, when parent server is esx+standalone
  • BUGFIX: (#19712) refresh data cause NPE, when client has unknown location
  • change: add schedule cli parameter -C <pcount>; set default pcount=1
  • BUGFIX: (#19812) GUI pop-up: The backup type 'NDMP' is not longer supported with this SEP sesam version
  • bugfix: correct cli notification switches for module (shortFields="M") and objectType (shortFields="o")
  • BUGFIX: (#19611) Avoid NPE when starting GUI in operator mode change: eliminate duplicate tag
  • bugfix: (#19768) Backport of commit ce7f112 - Could not handle notifications with an &amp

Backup

  • BUGFIX: (#19904) sbc.exe core-dumps for unknown reason. Problem: String was interpreted in printf. Solution: Avoid string interpretation
  • BUGFIX: (#19830) sm_sms_watch cored with exit code 0xc0000409. Problem: If sm_sms_watch checks the Si3 status during shutdown of Si3 service then the output complains about saveset information for every saveset. Solution: Check for proper end of result string
  • BUGFIX (#19612) - Hyper-V restore fails with FAILED. Method name is "ValidatePlannedSystem". Improved the sm_reformat_lis.py and adapted the vss module to get writer metadata of Hyper-V writer in the NOT file;
  • BUGFIX: (#19827) VSS Backup ends with 'HardlinkCleanProcessedGroup: Hardlink ... not found'. Problem: Processed hard link cannot be found in list. Solution: This is not a severe problem so go on with processing.
  • BUGFIX: (#19867) sbc.exe core-dumps upon specific file. Problem: The source with file name 'Merkblatt%201.8neu.pdf' could not be decoded with UTF-8 the trace message lead to core dump. Solution: Avoid printf format expansion for invalid text
  • BUGFIX: (#19857) VSS backup create warning with 'Volume is empty - nothing to do with it'
  • BUGFIX: (#19841) Exchange 2007 backup ends with 'sm_reformat_lis returned exit state 101'. format_prex() function is skipped if client is outdated;
  • BUGFIX: (#19843) Remote Access (CTRL/SM_SSH) does not send correct exit code sporadically. Problem: GetExitCodeProcess sometimes return wrong code (0). Solution: recheck exit code 0 using GetExitCodeThread
  • CHANGE: Use new Windows BSR Pro 1.5.142
  • BUGFIX: (#19832) Allow all backup levels for backup type 'vSphere'. There was no update of 'DB:task_types' for 'older' SEP sesam versions
  • BUGFIX: (#19582) Incremental backup of MSSQL DB stuck sporadically ROOT CAUSE: name of DB for mutex was incorrect, RESOLUTION: - get correct name of DB by checking which one is present wchar_t or char
  • BUGFIX: (#19748) MSSQL backup of named instance fails with 'DB_OpenItem: VDS::Create failed: x80770007'. Problem: Instance with name 'MSSQLSERVER_01' matched with 'MSSQLSERVER' and was therefore skipped. Solution: Improved check for 'MSSQLSERVER' instance.
  • BUGFIX: (#19731) - Hyper-V v2 backup with CSV fails with "Browsing failed with" and "Error: Unrecognized option: -c"
  • BUGFIX: (#19358) In case backup medium has been changed manually, which is still appendable: Do not try to read tape label, if one backup is already running.
  • BUGFIX: (#19146) BSR must support firewall port range and RDS (tape_server) setting. Problem: The firewall port range was not yet used up but the STPD returned: '425 Can't open data connection. WINSOCK: Address already in use. (0x2740,10048)'. Solution: Retry SData socket connect till TIME_WAIT is gone

Migration

  • NEW: (#19503) Migration: Use interface set in migration event for target server. (Read part is done locally on source server, so firewall shouldn't be a problem

Restore

  • BUGFIX: (#18745) After system state restore *_new files are not deleted when the computer reboots. Problem: Sometimes the system might not rename the specified file or directory successfully at restart. Solution: If the destination file or directory already exists and has to be replaced, the MOVEFILE_REPLACE_EXISTING flag also needs to be specified along with the MOVEFILE_DELAY_UNTIL_REBOOT flag when MoveFileEx() is called.
  • BUGFIX: (#19755) Connections to SEP sesam Server need long time during BSR restore with FW options set. Problem: If sbc restore processes data before callint last recv() with 0 (EOF) then communication ends with WSAECONNRESET (10054). Solution: Use SO_LINGER and DisconnectEx() only for SBC_COM or if restore data length was requested (BSR Pro restore)
  • BUGFIX: (#18976) BSR Pro Windows: Restore from DataStore with slow throughput 30 GB/h - improve with ReadCache for FTP and HTTP
  • BUGFIX: (#19786) Selective NetWare generation restore can fail with 'No parent directory for file'. Problem: SIDF needs parent directory for every selected file in every selected saveset FULL/DIFF/INCR. Solution: Add parent directories in 2nd BLOW_UP()
  • BUGFIX: (#19728) Restore of migrated backup stops with 'MTF_ERROR_BAD_ARCHIV'. Problem: A next tape was requested but sm_sms_watch did not check for migrated saveset. Solution: Check restore_results.saveset_used to get the media labels
  • BUGFIX (#19612) - Hyper-V restore fails with FAILED: "\533E17CD-CFFD-4994-B2A2-95DA2D0FDAA3\533E17CD-CFFD-4994-B2A2-95DA2D0FDAA3.vsv". Adjusted VM configuration file in case of restoring virtual machine that was backed up in "saved" state.
  • BUGFIX: (#19717) BSR windows recover impossible if backup server is Linux. Problem: SQL select statement with 'state=0'. Solution: Use 'state='0' for PostgreSQL
  • BUGFIX: (#19180) BSR-Restore doesn't work with Si3. Problem after reaching EOF the EOF state was not reset. Solution: Reset EOF state during rewind
  • BUGFIX: (#19655) sm_sbc_com doesn't find saveset if only migrated version exists. Problem: Select in result_lbls for original saveset failed. Solution: Use migrated saveset if original saveset does not exist
  • BUGFIX: (#19642) Domino generation restore skips needed saveset. Problem: Restore does only restore from recent FULL, DIFF and INCR saveset. Solution: Restore data from all selected savesets (complete FDI saveset chain)
  • BUGFIX: (#19647) Restore of last file from cpio saveset failed with: 'E010-RESTORE Restore failed. No data has been restored'. Problem: LIS without 2nd offset for last Lotus Notes file. Solution: Go on with cpio 'TRAILER' processing after last item processed

Media

BUGFIX: (#19628) Media init: Allow use of empty media, if there isn't any media from this pool in Sesam DB

Kernel

  • BUGFIX: (#19887) sm_qm shutdown should terminate all sub-processes. Problem: On Windows the sub-processes of sms and sm_java are not always ended with sm_qm shutdown. Solution: Since commit 421fb08 (#18938) Nov 25 2015 the TerminateJobObject() was not called for sm_qm shutdown

Disaster Interface

  • BUGFIX: (#19398) Changed the text of the disaster email according to the bugzilla call. Reinserted the restore command line and the loader/datastore information to the email.
  • BUGFIX: (#19398) Inserted additional information to the first parameter of the sm_disaster call to simplify the understanding of the parameters in the sm_disaster interface.

SMS

  • BUGFIX: (#19845) Windows: Datastore info file does not store the saveset information. Problem: Command line was shortened due to newlines. Solution: Avoid newlines by using URL decoded string
  • BUGFIX: (#19785) Shared drive on AIX reading from tape failed due to missing Filemark if tape was initialized on Windows written on tape before tape trailer. Problem: Windows dues not write FM after EOD. Solution: Write extra Filemark in TP_WriteTapeTrailer() and TP_WriteTapeTrailerForInit()

Installation

  • CHANGE: For code signing certificates Microsoft Windows will stop accepting SHA-1 signed code and SHA1 certificates that are time stamped after 1 January 2016. SHA-1 signed code time stamped by an RFC 3161 Time Stamp Authority before 1 January 2016 will be accepted until such time when Microsoft decides SHA-1 is vulnerable to pre-image attack. Solution: Dual signatures must be applied to the PE files EXEs and DLLs.
  • BUGFIX: (#19683) additionally sign the installer exe file with the SHA256
  • BUGFIX: (#19683) insert the new signature for the signing of the msi and the exe.
  • BUGFIX: (#19658) Installing bsr pro did not work for GUI package with client.
  • BUGFIX: (#19699) GUI icon needs admin right to be able to change the sm.ini value for the java_interpreter.
  • BUGFIX: (#19676) If sm_vfs executable is missing, do not check for the cbfs version file.
  • BUGFIX: (#19658) Correct the file names of all shortcut files on x86. Do not add additional .lnk to the shortcut files names.
  • BUGFIX: (#19092, #18403) insert the redistributable package for windows server 2008

Major fixes and changes v4.4.2.84

For a complete list of GIT change reports, see changes.

Installation/Update

  • BUGFIX: (#19676) Update from versions 4.4.1 or earlier hangs. Problem: If sm_vfs executable is missing the check for the cbfs version is not possible. Solution: Do not check for the cbfs version file if sm_vfs executable is not present.

Major fixes and changes v4.4.2.83

For a complete list of GIT change reports, see changes.

GUI

  • CHANGE: Added a tooltip text to the follow up label. Open internetbrowser, if click followup link at tooltip
  • BUGFIX: (#19614) Loader Type DIR-DISK SayFuse deletes slot allocation in case a second sayFuse loader is configured. Problem: table loaderdevice will be cleared completely, when set devices of one loader. Solution: filter deletion by current loader
  • BUGFIX: media cannot be edited. Problem: media.drive is empty. Solution: add empty check for media.drive.id at update method
  • BUGFIX: the prio of an immediate start event was not handled correctly
  • BUGFIX: topology tree is empty, when hwdrive has unknown client
  • CHANGE: (#19497) improve error message, when try to remove datastore with replication tasks; set correct font and rename ok/cancel button of message dialog
  • CHANGE: (#19497) improve error message, when try to remove datastore with replication tasks. Added important warning with cancel action button, when delete datastore, which has si3 replication tasks
  • bugfix: test 'testYearlyRelative' failed
  • BUGFIX: (#19603) SEP sesam license timed out due to wrong installation date (e.g. 1976-05-14). Problem: complete params_fix was overwritten, when change default settings. Solution: don't update the readonly field
  • bugfix: CLI dir client used wrong output format. Problem: 2 blanks after type ("/xyz/" dv - - - - 0 - ,)
  • BUGFIX: (#11988) Check if a migration taskname is already in use when renaming a task
  • BUGFIX: (#16522) prevent create datastore via cli with lowwatermark, which is greater than highwatermark (pos values); improve cli error messages, if cli uses invalid values for highwatermark/lowwatermark/capacity
  • CHANGE: (#19579) improve error message, when create new drive with drivenum, which already exist
  • BUGFIX: (#19343) The sorting-order of the tasks in a taskgroup changed when a task was assigned to a schedule in the task dialog
  • BUGFIX: (#17748) Backup will Fail with "InvalidLogin" if password for vcenter contains special char "ß". Problem: Password enryption/decryption won´t work with umlauts/special chars. Solution: convert password to UTF8 after decrypt password and use it for VM vSphere access. When encrypt password with new saveset (VMDK saveset), convert it back to charset 'ISO-8859-1'
  • BUGFIX: (#19578) Drive cannot be deleted at drive properties. Problem: Drivenum is empty at drivedialog. Solution: Initialize max drivenum filter before set drivenum
  • BUGFIX: (#18859) Import of license by GUI failed (Read InputStream and unzipped file with UTF-8 encoding)
  • BUGFIX: (#19574) It is not possible to add a schedule to a task. Problem: NPE, when trying to add schedule to task. Solution: add NPE check
  • BUGFIX: (#19571) GUI updates EOL of running backup, which can result in wrong EOL. Problem: result table will be updated, when leaving the dialog with button 'ok'. Solution: add write lock during active backup at result dialog for fields 'EOL', 'user_comment' or 'locked state'
  • BUGFIX: (#19558) modify existing taskevent at taskdialog cause adding a new one
  • BUGFIX: (#19557) VMware single file restore does not work sporadically due to multiple thread calls. Problem: command 'umount' will be executed, after mount vmdk at restore wizard, when mount VMDK. Solution: remove dismounter at server side, which clean mount session after mount
  • BUGFIX: (#19558) GUI-schedule/task handling is wrong - the task is (again and again) added to ALL selected schedules every time APPLY and OK is clicked. Problem: insert event will be done two times, after press button 'apply' and 'ok'. Solution: mark inserted event at internal event table for prevent further inserts of same event
  • CHANGE: (#19515) optimize error management; error 'VM datacenter,datastore,... not found' appears at restore wizard, although connection to vcenter server failed;
  • BUGFIX: (#19515) clients with OS 'ESX Server' are not available as target server, when restoring a VM
  • BUGFIX: (#19525) restore VM from vcenter server to esx server does not work. Problem: wrong network.addresstype 'assigned' at network spec (MAC address assigned by VirtualCenter). Solution: set network.addresstype to 'generated' (Automatically generated MAC address)
  • BUGFIX: (#18676) Allow drive number other than 99 if flag max_drives is set in the defaults table example: key=max_drives, user_name=sesam, value=999
  • BUGFIX: (#19514) Linux clients couldn't be selected as data mover for ESX, during client creation
  • BUGFIX: (#19538) Modify datastore deletes path from DB. Problem: datastore path will be replaced with empty string, when modifying datastore with cli. Solution: init datastore path variable with null instead of empty string ""
  • BUGFIX: (#19541) Avoid NPE when opening results of a backup group
  • BUGFIX: (#19515) Avoid NPE when updating the VMPanelData
  • BUGFIX: (#19518) Avoid NPE when filling the EsxModel
  • BUGFIX: (#19518) ESXi cannot be selected for vSphere restore as target (fix in ClientsCache)
  • BUGFIX: (#19507) Restart Backups view also showed successful backups
  • BUGFIX: (#19515) vSphere Restore is not working if vCenter is unreachable; (no VSphereItemListener active)
  • BUGFIX: (#15284) Changed text "Device (SCSI-ID)" for loaders
  • BUGFIX: (#19504) Source and target drive for replication have to be selected
  • BUGFIX: (#19494) After renaming a backuptask, a restore of a finished backup doesn't work due to NPE. Problem: try to generate rTask name with task, which does not exist anymore. Solution: get task name from results
  • BUGFIX: (#18306) Use calendar calculation to avoid issues with summer/winter time
  • bugfix: create client via CLI use always locationId=0
  • BUGFIX: (#19408) The preselected location was not correct in case of sublocation with same name
  • bugfix: (#19062) added git hook to maven build

Backup

  • BUGFIX: (#19627) Remote command over 'ctrl': Return correct exit code (error), if command doesn't terminate correctly (if killed by signal for example). Check of exit code has been done by using 'WEXITSTATUS()' macro without checking 'WIFEXITED()' 1st.
  • BUGFIX: (#19608) BSR-Backup fails because of read-errors. Problem: Size given in boot block of volumes (either GPT or MBR) does not match with range on disk. Solution: Set backup parameter 'ignored_read_errors' to 'YES' to ignore such error messages and to go on with backup.
  • BUGFIX: (#19619) BSR Pro doesn't show single started BSR backups. Problem: BSR tasks started without group have session_id set to 'single' since commit 1551afe: (#19520) BSR backup on Windows: Do not set 'DB:results.session_id' to 'saveset'. Solution: Select savesets with session_id 'single' too.
  • BUGFIX: (#19852) Incremental backup of MSSQL DB stuck sporadically
  • BUGFIX: (#19590) SAP hana backups hang if connection from sm_sbc_com cannot be established. Problem: STPD Data first socket connect returned WSAEADDRINUSE and 2nd attempt failed with WSAEADDRNOTAVAIL. Solution: Execute 2nd connect with same socket settings in case of WSAEADDRINUSE
  • BUGFIX: (#19598) Sharepoint backup fails with 'WIN32 API error: 3 - The system cannot find the path specified'. WORKAROUND: function sleep() is moved up to DB API level.
  • BUGFIX: (#19529) The BSR Pro service should be started if the service is not running. Problem: Backup fails if service is not running. Solution: Check state of service and start if not running.
  • BUGFIX: (#19528) Do not write an entry into 'DB:result_lbls' during EOM for task groups, but only for 'real' backups
  • BUGFIX: (#19520) 'BSR Windows': Do not try to insert into 'DB:result_lbls', if backup is started by 'sm_backup' (which has already made 1st insert
  • BUGFIX: (#19520) BSR backup on Windows: Do not set 'DB:results.session_id' to 'saveset', because this overwrites reference to task group results entry.
  • BUGFIX: The virtual machine can not be backed up because the setting for incremental backups is enabled and there is a connection with iSCSI disks. Unset WMI property for each VM allowing to make INCR backup;
  • BUGFIX: (#19460) Lotus Notes DIFF/INCR backup ends with 'Error: Option -T requires an argument.' if previous backup was a FULL backup. Problem: Incr time was set to empty string. Solution: Do not overwrite incr time from FULL if DIFF/INCR not found.
  • BUGFIX: (#19175) Notes backup adds / to path and as such files are not restorable from restore wizard. Problem: Path is set with beginning '\'. Solution: Avoid prefix of '\' to valid path.

Restore

  • BUGFIX: (#19616) Selective restore of Novell iFolder object fails with 'Cannot find matching LIS line for item'. Problem: The root entry '/' was not retrieved in the 2nd BLOW_UP. Solution: Do not skip '/' for root entry
  • bugfix: (#19540) VMware CBT restore does not connect to VMDK anymore
  • bugfix: (#18655) BSR Pro Windows: Restore from DataStore with slow throughput 6 GB/h. Problem: Saveset file was reopened if segment file with tape label was accessible. Solution: Use remembered file handle for read even if tape label is given.
  • bugfix: (#19499) VADP restore via CBT sporadically fails with "cbt_set_item: incorrect vmdk block format:"
  • BUGFIX: (#19291) To obtain all files needed for a Hyper-V restore the lis and the not file are compared. Problem: If the files are written in different case style the matching did not work. Solution: Change the matching to case insensitive.
  • BUGFIX: (#19291) CSV HyperV VM restore fails with '_Vss_Backup_Component_.xml with error WIN32 API error: 2 - Das System kann die angegebene Datei nicht finden.' Two issues are fixed: 1. case-sensitive search for "renamed" rule; 2. xml files of VM have "renamed" rules as well;
  • bugfix: (#18655) BSR Pro Windows: Restore from DataStore with slow throughput 6 GB/h. Problem: Saveset file is reopened for every data request. Solution: Keep read file handle open and reuse it.
  • BUGFIX: (#19550) Selective restore to original destination of a migrated MSSQL backup uses first database in list. Problem: LIS file of migrated saveset for MS SQLServer has wrong offsets due to blocking with 512 Bytes. Solution: Check the SFMB offset_to_data and use it for backup and migration.
  • BUGFIX: (#19550) Selective restore to original destination of a migrated MSSQL backup uses first database in list. Problem: The function MAP_SSL_FILE() does not take care about enclosing TAPE, SSET, VOLB and ESET. Solution: Use l_BLOW_UP_SLL() to get enclosing meta data streams TAPE, SSET, VOLB and ESET for selected MS SQLServer databases
  • BUGFIX: (#19539) Selective restore of MSSQL DB uses wrong object if another DB exists with a name which is a part of the selected DB name. Problem: The second database name was a part of the selected one - for instance 'DB_1' is a prefix of 'DB_12' - and the restore algorithm added this object during check for parent directories. Solution: Check for parent directories with ending '/'

License

  • NEW: License keys for 4.4.3: 'Volume_L2_TB', 'Volume_L3_TB', 'JIRA', 'RHEV_SERVER', 'RHEV_SERVER_SEATS' and 'NDMP_Extension'
  • BUGFIX: (#19422) License encryption invalid if first sign is space. Problem: Customer inserted space chars before license line. Solution: Trim the line (skip space)
  • BUGFIX: (#18866) License check counts too many mailboxes in case of Exchange DAG. Problem: Mailboxes of Exchange Server DAG members are counted too. Solution: WORKAROUND: Count only max(mailboxes) do not cumulate.
  • BUGFIX: (#18004) vCenter license is calculated incorrectly. Problem: In version 4.4.2.78 Datatcenter are counted instead of vCenter. Solution: Count every '* vCenter ' line but ignore duplicates in output of 'sm_cmd list vSphere'

Kernel

  • BUGFIX: (#19372) sm_vfs updated to support cbfs v6
  • BUGFIX: (#18710) Encrypt/decrypt of passwords for sbc doesn't work with 'umlaut'. Problem: Non ASCII chars are not decrypted properly. Since commit c821def (May 11 2011) the compiler warnings have been resolved but now chars with value higher SCHAR_MAX (127) lead to prefix Byte 0xFF. Solution: Set prefix Byte to 0x00 like old algo did.
  • CHANGE: Display socket address IPv6 and IPv4 for target RDS host. Use first IPv4 or IPv6 address
  • BUGFIX: (#19454) To prevent the log file from getting too large in case the qm is not able to start, shutdown the qm if error 'error: WIN32 API error: 6 - The handle is invalid' occurs.
  • BUGFIX: (#19506) SMS log level stays at 6 after EOM at restore. Reset log level after EOM handling.
  • NEW: sbc_vss.exe enhancement - new options -set enableincrementalbackup|disableincrementalbackup
  • BUGFIX: (#19492) Extend 'DB:media_results.msg' to 1024 chars, to match message length in 'DB:data_stores'

Installation

  • BUGFIX: (#19138, 19491) changed the creation of the shortcuts to also support Windows 10. Use MSDN function for the creation now.
  • BUGFIX: (#19372) copy the cbfs driver version 6 in the installation package.
  • BUGFIX: (#19138, 19491) change the creation of the shortcuts to also support Windows 10. Use CBFS driver version 6 instead of version 5. Uninstall old CBFS driver 5 before installation of CBFS driver 6.
  • BUGFIX: (#19454, #19333, #19292) the install of the new service seems to produce an error in some cases. Solution: Only update the service binary file in case it is the old service binary file sm_main.exe, otherwise leave the service unchanged.

Major fixes and changes v4.4.2.78

For a complete list of GIT change reports, see changes.

GUI

  • BUGFIX: (#19442) When deleting a datastore NO warning message appeared
  • BUGFIX: (#19355) Avoid NPE when creating a new migration_event and select the "empty" value for task name
  • BUGFIX: (#17713) With parallel migration, not all jobs are displayed under Current Messages/Performance
  • BUGFIX: (#19393) sm_cmd does not create unique ID's if option 'migrate' is started twice. Problem: insert migrationtask failed due to PK name not unique. Solution: get unique id via command 'sm_info get_unique_id'
  • BUGFIX: (#19404) View of "replication tasks" and "migration tasks" show wrong values
  • BUGFIX: (#19368) GUI server does not track user action; introduce RequestAppender in this version; this is done by backport of #19240 - add request logger
  • BUGFIX: (#19383) STPD port for RDS can be set, but not be unset
  • BUGFIX: (#19349) Reverted that the context menu in the saveset is column based (keep user habit).

Backup

  • CHANGE: insert the new SEP sesam BSR Pro version 10.0.120
  • BUGFIX: (#19291) CSV HyperV VM restore fails with '_Vss_Backup_Component_.xml with error WIN32 API error: 2 - Das System kann die angegebene Datei nicht finden.' Implemented the restore with renaming mechanism for Hyper-V according to rules specified in xml backup document.
  • BUGFIX: (#19434) Lotus Notes INCR backup may skip active transaction log cos since time refers to last DIFF backup. Problem: For Lotus Notes the INCR time must be set depending on the level DIFF/INCR but the based_on for the restore chain must be set to previous backup DIFF or INCR. Solution: Execute 2 select statements to get the incr_time and the based_on separately
  • BUGFIX: (#16028) A Lotus Notes diff backup with empty dataset generates an error 'no backup source found'. If no data is retrieved for a DIFF backup then return with success
  • BUGFIX: (#19434) Lotus Notes INCR backup may skip active transaction log cos since time refers to last DIFF backup. Now Lotus Notes INCR backup refers only to last FULL/INCR and DIFF backup to FULL/DIFF.
  • CHANGE: (#19429) Include VDDK 5.5.4 as default instead of 5.5.3. VDDK 5.5.4 is necessary for vSphere6 to use SAN and HOTADD and VMware just added the support for it there.
  • BUGFIX (#19335) - Wrong source is determined for Hyper-V 3 VMs. The filter "where Caption ='Virtual Computer'" to get list of VM's by using WSQL query of WMI is localized - used another one by field "Description".
  • BUGFIX: (#18523) if no tape is in drive the drive was not set as encryption capable. Now if no tape is in drive the encryption capability of the drive is check anyway.
  • BUGFIX: (#19146) BSR must support firewall port range and RDS (tape_server) setting. Problem: The Sesam server interface was accidentally returned in szGET_SERVER_INTERFACE(). Resolution: Return correct RDS interface.
  • BUGFIX: (#19403) Windows: MaxDB backup failed due to path with blanks, e.g. 'C:\Program Files\SEPsesam\bin\sesam\sbc'. Enclose program in "
  • bugfix: (#19390) ME13 Restore to generated RDB - Error - RESTORE STATUS: E012-RESTORE Restore was not successful. Error: Unable to cast object of type 'Microsoft.Exchange.Data.Directory.Recipient.RequestStatus' to type 'System.String'. Make cast to RequestStatus.
  • BUGFIX: (#19132) Do not strip leading '//' (UNC path) from datastore path during saveset mount over VFS
  • NEW: (#19375) Create a unique media label for inquire operation from Oracle/RMAN, to allow parallel restore from Sesam data stores. TODO: For MML 2.0 a shared access flag can be set on 'libobk.so' side to allow multiple read operations from one medium.
  • BUGFIX: (#19163) sbcln.dll cannot handle special character ß. When calling DB_PrepareItem() the wide char string must be converted back with locale encoding not with UTF-8

Major fixes and changes v4.4.2.76

For a complete list of GIT change reports, see changes.

GUI

  • BUGFIX: (#19019) When creating a new vSphere backup task, the hourglass runs. The creation of a new vSphere backup task works now.
  • BUGFIX: (#19379) ClassCastException when try to start migration, add migration_event
  • BUGFIX: (#19353) sm_cmd remove fails for several objects. Problem: cli tries to execute 'remove(Object...) method' which does not exist. Solution: scan type Object and run method 'remove' with param 'string/long'
  • BUGFIX: (#11459) When starting gui the first time, cyclic flag is off in the current messages pane. This will be saved with profile now.
  • BUGFIX: (#19349) EOL of saveset cannot be changed: Misleading use the same symbols for EOL and Locked leads to suggested switching possibility of Locked in the EOL column. Change is made on the fly - no OK button needed.
  • BUGFIX: (#19350) Migration task: some filter settings are not saved. Problem: 'cfdig' comboboxe types will not be set, when reopen migration dialog; selected task/group get lost, when change migration task
  • BUGFIX: (#19343) The sorting order of tasks in a group changed when a schedule assignment was changed
  • BUGFIX: (#19337) Avoid ClassCastException when trying to add a user defined media type
  • BUGFIX: (#19300) Different amount of data displayed in Select Task and Save/Restore in the Restore Wizard
  • BUGFIX: (#19329) Filter by 'error','warning','failed' at media by status does not work. Problem: several failure states exist ('media_failed' {"f"}, 'media_action_failed' {'x'}), both render same string 'failure' for column 'media_state'. Solution: add new filter 'media_action_failed'
  • BUGFIX: (#19326) Tolerant treating of entries with false second values e.g. sbc_start=20151113110999
  • BUGFIX: (#19176) NullPointerException when expanding gui window to full screen or back to normal size
  • BUGFIX: (#19306) State filter of migration by status tree table does not work
  • BUGFIX: (#19321) Avoid error Invalid Value: EXCLUSIVE_BUSY in HwDrivePerfFetcher#fillPerformance
  • BUGFIX: (#19274) sbc output not visible during restore over mount saveset problem: only error output of command was streamed, add stream output of command
  • BUGFIX: (#19165) Suppress warning message about replicated saveset
  • BUGFIX: (#19270) shows "resource" as message in cancel dialog
  • bugfix: (#19194) Creating a loader by clicking 'apply' -> 'ok' creates two loaders
  • BUGFIX: (#19244) Add a task to a schedule over the task properties via "Apply" and "OK" - genearates two task_events instead of one
  • BUGFIX: (#19245) Added missing german translations
  • BUGFIX: (#19264) New Replication Task couldn't be created. Remove newday event failed. Problem: has still dependencies to table 'terms', 'calsheets'. migration task columns submit_flag, migrated_flag will be filled with "true/false" instead of "1/0"
  • BUGFIX: (#19252) No terms entry is created for additional newday event
  • BUGFIX: (#19249) Mediapool change of a schedule runs into 'Database is busy' (start executeSMDailySchedule via invokeLater)
  • BUGFIX: (#19246) The last view timestamp was not updated anymore
  • BUGFIX: (#19116) Populate missing data in rows Last Full Backup and Last Successful Run
  • bugfix: (#19215) Saving dates in a view saves them absolute and not relative. Problem: newday date exists two times at result date collection, date offset was shifted for one position
  • bugfix: (#18877) Prevent files which are too large from being read (>50 MB). Instead show message: "This file is too large to be displayed, please analyse file in the original location at XXX"
  • BUGFIX: (#19228) GUI offers "SEP sesam Recovery Pro for Exchange" for encrypted savesets. Encrypted save set cannot be mounted so this function is not available anymore.
  • BUGFIX: (#19221) After a scheduled media init the "new" initialized tape will be unloaded and the previous tape will be uses instead. Problem: default set media_pools.close_on_init='0', when create, update schedule and set media_pools.close_on_init='y', when create/update mediapool and field is uninitialized. Solution: if panel 'close_options' is invisible, set null
  • BUGFIX: (#18135) Clients were always created under first occurrence location with the same name. Now the client is created under the correct location.
  • BUGFIX: (#4676) User authentication and authorization: Turned off the any authority checking in V4_4_2
  • BUGFIX: (#19197) Added a call of deactivate_uac in the permission management frame
  • BUGFIX: (#19162) migration_results show childs as flat items and also under parents
  • BUGFIX: (#11322) Profile will be saved, but cannot be set as default. Problem: default profile cannot be saved, if no default profile exist at DB
  • BUGFIX: (#18448) Job state tabs are empty after sm_newday. Problem: From/to combobox will not be refreshed after newday event for restore, migration, media jobs table)
  • CHANGE: (#19133) Same drive_type must be set if a new drive is added to a drive_group which has already a configured drive)
  • BUGFIX: (#19174) No tree view for restore results
  • BUGFIX: (#18519) Changing the "To" (recipient) in the Send Mail option has no effect
  • BUGFIX: (#19077) 'Auto Update' arrows no longer change the color of the one arrow to the other arrow
  • CHANGE: (#15560) Select the preselected save set in restore wizard if started from result table
  • change: add params '-B' (begindate) '-E' (enddate) for filter output of 'list restart' by from/to date like 'list results'
  • BUGFIX: (#19156) sm_cmd list restart doesn't work anymore
  • BUGFIX: (#19151) INC mount with RPEX fails, because "sm_data_store -o gen" parameter is missing
  • bugfix: (#19108) Region parameter de_by faultly sets GUI language to German. Bavarian GUI works again.
  • BUGFIX: (#19140) Avoid NullPointerException when creating a new migration task in schedules
  • BUGFIX: (#19141) Avoid a possible InvocationTargetException when starting the gui
  • BUGFIX: (#19137) Avoid stack trace after selecting the "remind me later" option
  • BUGFIX: (#19129) Avoid possible NullPointerException when choosing a media pool in the Task By Clients dialog
  • BUGFIX: (#18981) Insert unique constraint over hw_drives.client_id and device
  • BUGFIX: (#18872) Avoid extra exception in 'Send Test Email' if you do not specify user/PW
  • BUGFIX: (#19103) Schedules with execution "once" are saved with daily execution. Problem: save function works, init schedule dialog with once schedule failed. Solution: add missing case 'once' for set panel 'once'
  • BUGFIX: Correct switching which depends on backup type in VM Row, IMAP Row and Mount Row
  • BUGFIX: (#15915) Show overwriting VM warning dialog if the original Xen VM will be deleted
  • BUGFIX: (#19099) No port/slot range possible in loader actions
  • BUGFIX: (#19095) Restart failed, because backup sets wrong start time at generated schedule (always start_time=01:00)
  • BUGFIX: (#19065) Could not save restore task if media isn't in loader
  • BUGFIX: (#18904) Changing operating system and platform on a new client causes a NullPointerException
  • BUGFIX: (#19066) Wrong start time set for "sm_cmd restart". Problem: generated schedule set null for starttime, startdate. Solution: set current time for schedule starttime/date
  • change: ExceptionDialog: Append the content of the text area panel to the stack trace only if is not blank
  • BUGFIX: (#19056) Empty stack trace window if exception occured
  • BUGFIX: (#19069) Remove newday event not possible. Problem: remove newday event without id '0' will be removed from schedule tree, but DB will not be updated. Solution: add newday event at remove event method and invoke 'remove' method at server newday event dao.
  • BUGFIX: (#19061) JobState/Filter: No previous day could be selected (handle format "yyyy-MM-dd HH:mm:ss 0" as a valid date; improve logging of invalid values in the gui server log)
  • BUGFIX: (#19051) Could not delete a client after update 4.4.1->4.4.2
  • BUGFIX: (#19050) VMware FULL CBT backup without mode 'image' (VSPHERE:) does not create CBT files problem: get changed blocks with wrong VMDK format. solution: get changed blocks via source VMDK path without snapshot ID
  • BUGFIX: (#19049) Java NullPointerException exception when opening loader contents window after failed archive adjustment Problem: fields loaderContent.fromObject, loaderContent.toObject were null
  • BUGFIX: (#18663) GUI Restore Wizard does not allow to restore a BSR-Backup saveset (e.g. to dump file) Problem: no external savesets will be shown, when continue without this option, click 'back' and enable it (show external savesets)
  • change: add gui properties for node 'commandevent'
  • BUGFIX: (#19042) Suppress flag isn't saved during creation of a new command/migration/media event
  • BUGFIX: (#19037) sm_cmd sql throws NullPointerException if result is empty. Solution: prevent show NullPointerException, if sql query has no output, show correct sql error message as retval, if sql query is invalid (no such column, no such column, ...)
  • BUGFIX: (#19032) Restart a task via IMMEDIATE START produces an almost empty window; add NullPointerException check for mediapool
  • BUGFIX: (#18747) The wrong drivegroup name template is used if a new datastore is created.
  • BUGFIX: (#19022) Set "Copy migrated savesets again" and "Parallel Migration" as a default
  • BUGFIX: (#18602) Corrected popup license message from Sep sesam Server
  • BUGFIX: (#19011) Avoid shrinking the restore wizard too much so that all controls stay always visible
  • BUGFIX: (#18953) Additionally check whether a mail action is supported on the current platform. Start local mailer in a timer controlled thread to avoid freezing the gui
  • NEW: (#18192) Performance tuning in UI server: Optional GUIServer parameter to start the web interface; Argument "--startWebInterface" or in sm.ini: startWebInterface=false; default is FALSE
  • BUGFIX: (#18993) Avoid stack trace when customer is trying to send a log file by e-mail. The accounts dialog did not close when it was started from the send mail button of a result protocol
  • BUGFIX: (#18985) Drives properties from the DataStore were not displayed if the panel was set to a very small size
  • BUGFIX: (#18852) Open property of backup result and close dialog with eol changes show message 'You are about to change the EOL to an already expired point in time.'
  • new: (#18928) REST interface 'sep/api/info/cliService/executeCli' for execute CLI command line execute cli command with curl failed, correct curl syntax: curl -X POST -d "['list', 'clients']" "http://beifus:11409/sep/api/cliService/executeCli". execute cli via rest(curl) deliver no data, request param must use template "..."
  • bugfix: action 'start' does not exist for cli action 'start inventory'
  • BUGFIX: (#18980) Archive adjustment via cli couldn't be started. Problem: Loaderinventory service not found Conflicts: sep-server/src/main/java/de/sep/sesam/gui/server/communication/RestHandler.java

VMware Vsphere

  • CHANGE: add NullPointerException check, when check recover param contains 'vmdk'
  • bugfix: VM restore with recover=addvmdk failed due to error 'VM not found'
  • BUGFIX: (#19290) VADP snapshot is not deleted in case of failed restore
  • BUGFIX: (#19281) Relocation at VADP restore uses wrong VMDK. Problem: VMDK order of param '-R reloc' is wrong and problem with param '-R VMDK1==::==[<datastore>] VMDK0' during parsing. Solution: exchange of source/target vmdk (-R <source VMDK>==::==<target VMDK>)
  • BUGFIX: (#18373) NullPointerException during diff VM backup. Problem: source VMDK path contains snapshot id. Solution: use only source VMDKs as key for VMDK map
  • BUGFIX: (#19238) Restore of VM config fails with: message=sm_sbc_com-E007-SBC_COM Missing mandatory save set identifier. Problem: restore vm config try to restore VMDK data, which does not exist. Solution: set 'recover=rest.conf.vmdk' as default for restore VM config, disable VMDK modes at VM recover option panel
  • bugfix: stop VM backup with correct error msg, if VM snapshot will be created with 'null' value, create snapshot name again and do null check, before stop procedure
  • BUGFIX: (#19211) vSphere backup fails with "Exception caught while invoking method: Logout". Problem: vsphere method 'logout' cause Exception 'NotAuthentificated'. Solution: catch exception and prevent log stacktrace
  • bugfix: run vmaccess with param -a 'vijava=false' cause lots of errors and will fail. Problem: disable vijava use directly vcenter server without api for vsphere server communication. Solution: stop vm operation with error 'unsupported param', when param vijava=false
  • bugfix: VM full backup with excluded VMDKs failes. Problem: commit: 'e8ad114' cause empty root vmdk data of VM, when create cbt config file
  • bugfix: run diff backup of VM with excluded VMDK failed. Problem: excluded VMDK will be used to generate CBT file, but VMDK path does not exist
  • bugfix: (#19134) Restore VM failed with error: sbc-1067: Error: XBSA Call BSAGetObject failed. Problem: tape server param '-S' has several ports separated by comma and will be cutted for command 'sm_vmdk_interface openjob'. Solution: remove separator ',' and use only newline for parse command output
  • BUGFIX: (#18737) diff VM backup failed due to retrieve VMDK via VMDK path for calculate changes blocks does not exist; get VMDK via drive position
  • BUGFIX: (#19114) start of VM backup with param '-l f' (full) failed. Problem: does not work with lower letters 'cfdi'. Solution: use case insensitive for set fdi type
  • bugfix: VMDK source path is missing at VM backup, if VMDK was created under snapshot
  • BUGFIX: (#19084) VM Restore to different Datastore structure doesn't work. Problem: create CBT file at the end cause Indexoutofboundexception, when get VMDK with snapshot id, which does not exist at root. Solution: if root VMDK does not exist anymore, use snapshot VMDK as source/target VMDK
  • BUGFIX: (#19093) VM diff backup will be set to level 'FULL', if VM has VMDK from different datastores Problem: VMDK path equals between two VMDKs, which are placed under different datastores Solution: Add datastore + vmdk path for check, if location of VMDK has been changed
  • BUGFIX: (#19038) sbc_vadp creates wrong CBT files if VM has more than one disk
  • BUGFIX: (#18737) vSphere diff and incr backups hang at "Get changeID" for the second vmdk; set correct VMDK for create CBT file
  • BUGFIX: (#18737) VM diff backup failed, if VMDK path contains snapshot id (vmdk3_000005-000001.vmdk)
  • BUGFIX: (#17158) GUI does not remember credentials during browse of vSphere, if user name or password was wrong at first request

Backup

  • BUGFIX: (#19384) 'all' backup fails on EFI systems. Exclude BCD files for all backup types
  • BUGFIX: (#19382) Exchange2013 backup with granularity=none and server=none fails with: 'The user name or password is incorrect'. Use local snap-in Microsoft.Exchange.Management.PowerShell.SnapIn instead Remote Management Shell to load Exchange PowerShell cmdlets in the Exchange 2013 env
  • BUGFIX: (#16745) Zarafa backup: Check return code from 'zarafa-backup' command and set backup state to 'Error', if at least one export has failed
  • bugfix: (#18997) Many trace messages during system_state backup. Last commit added 'PackArchiveBlock: Remember next file [...]' trace line
  • BUGFIX: (#19307) VSS snapshot fails if foreign snapshot provider is installed. Use default Microsoft Software provider.
  • BUGFIX: (#19293) System state - Missing important items. To process all files for the 'VSS Metadata Store Writer' with path set to 'C:\Windows\Vss\Writers' and filespec '*.xml' the RETRIEVE_PATTERN_RECURSIVE flag for iRetrieveItem() must be used
  • BUGFIX: (#19293) System state - Missing important items. Used the custom "Performance Counter Writer" instead original one on the Windows 2012 (R2)
  • BUGFIX: (#19159) - Windows Dedup-ChunkStore isn't excluded during backup. Implemented the exclusion logic for writers.
  • BUGFIX: (#19293) System state - Missing important items. In case of system_state backup sometimes a new directory is implictely prcocessed via given file, these files must be processed after pending hard links.
  • BUGFIX: (#19272) Exchange backup truncates transaction logs in case of VSS error during backup. Set the error code even the GetLastError() returns 0 in the DB_ItemFinishStatus().
  • BUGFIX: (#19214) System state - Missing important items. ASR writer original metadata is changed - some adaptation was needed for system_state.
  • BUGFIX: (#19012) System state backup fails on EFI Systems (fixed exclude pattern issue)
  • bugfix: (#18820) pagefile.sys must not be backed up during system_state backup. The default exclude patterns were not involved during the system_state.
  • BUGFIX: (#18997) Many trace messages during system_state backup. Avoid trace message in log level 0 'Partial hardlink ...' and 'HARDLINK_STATE_PARTIALLY_PROCESSED ...'
  • BUGFIX: (#19193) sbc hangs if host is unreachable. Since commit 78fd6f6 the BSAInit() sets up the communication so now if BSAInit() fails the GEN interface must be released
  • BUGFIX: (#18828) BSR backup fails with: Operation failed: BSR PRO unknown return code (0XFFFFFFFD). After calling the external BSR Windows trace function the error codes are set to 122 - Remember old values and reset after external trace function
  • BUGFIX: (#18828) BSR backup fails with: Operation failed: BSR PRO unknown return code (0XFFFFFFFD). Call WSAGetLastError() immediate after connect() and use the return value cos a further call may return error 122 if BSR external trace function is called
  • NEW: Accept Windows 10 as supported operting system and do not show the warning anymore.
  • BUGFIX: (#19055) - HyperV v3 backup with CSV fails with 'No volume could be determined'. Issue with transforming HYPERV:/Online/Vm - the wrong one is "VSS:/Microsoft Hyper-V VSS Writer/*/Online/Vm", the correct one is "VSS:/Microsoft Hyper-V VSS Writer/Online/Vm"
  • BUGFIX: (#19097) Oracle RMAN backup failed because same backup saveset ID (bkfilnam) is used again for control file backup. If saveset id is not unique then return sbt error 'SBTOPEEX 7001 : backup file already exists (write only)' to enforce RMAN to use another saveset id
  • BUGFIX: (#19039) ME10 Warning/Error - sm_reformat_lis returned exit state 103 - format metadata into "pretty" xml in the NOT file
  • BUGFIX: (#19030) Exchange META file couldn't be created if saveset encryption or compression is switched on. METADATA is printing out into NOT file with further process in the sm_exchange.py.
  • BUGFIX: (#18882) sob calls 'sbc_com_interface' sporadically without or wrong values. The problem was that the sm_sbc_com backup process needed too much time to check for semaphores on the server side. In case of a Data Store no semaphores must be checked.
  • BUGFIX: (#18966) STPD core-dumps during MSSQL backup - Support long paths in Logical names
  • BUGFIX: (#18966) STPD core-dumps during MSSQL backup - adding #ifdef to avoid Logical names retrieve
  • BUGFIX: (#19018) Restore of System_state backup does not show folders and files in Select Tree. In this case the metafiles must not be skipped from LIS file
  • BUGFIX: (#19016) HyperV v2 backup with CSV fails with: IsVolumeSupported returns: FAILED.] - VSS_E_PROVIDER_NOT_REGISTERED. Allow to use old client 4.4.1 together with SEP sesam server 4.4.2: USe old Backup prefix and source adaption
  • BUGFIX: (#18997) many trace messages during system_state backup. Avoid trace message 'PackArchiveBlock: filename not filled so set iRet = MTF_MORE_BLOCKS'

Restore

  • BUGFIX: (#19325) Zarafa: restore fails if mail subject starts with a double quote ". Get SEL line item type via positioning to last quote
  • BUGFIX: (#18975) sm_lis_items does not always create lst and lsl correctly in case of gen mode. Last item of generation LIS was not added to LST and LSL files.
  • BUGFIX: (#19338) Zarafa restore of complete mailbox from FULL saveset failed with ERROR: Cannot find matching LIS line for item ["IMAP:rs@sep.de/" ...]. Get all folders of selected mailbox
  • BUGFIX: (#19232) Zarafa: Generation restore does not work - result 'Empty saveset'. If some subject line contains " then the item type parsing may fail. Get item type with strrchr()
  • BUGFIX: (#19327) Generation restore fails with 'Recordkey XX was not found' wrong record key in msl file. In case of generation restore the SG LIS file format has extra saveset and line counter
  • bugfix: (#19303) Check for valid date/time format of sbc start time and use current time if not valid to avoid core dump in 'sm_sms_backup'
  • BUGFIX: (#19179) Restore fails with 'stpd fails while searching for media label in datastore path'. Do not search for saveset file, if no tape label is set in 'RETR' (restore) call and data saved on a Si3 deduplication store. Doesn't make sense here.
  • BUGFIX: (#19193) sbc hangs if host is unreachable. Since commit 78fd6f6 the BSAInit() sets up the communication so now if BSAInit() fails the GEN interface must be released
  • BUGFIX: (#19089) Restore of UNC path does not show folders and files in Select Tree Tab in Restore Wizard. Remove leading '/' for UNC path in LST and LSL file. Adapt logic in module sm_restore.
  • BUGFIX: (#19075) Restore of migrated savesets not possible. Create original LST file if it does not yet exist
  • BUGFIX: (#19033) Selective restore of a large amount of files produces a huge log file. Avoid to write every item into log file

Restore via Mount

  • BUGFIX: (#19282) Mounted generation saveset via sm_vfs does not show content in all cases. enhance lsl -> lslc conversion; fix problem with multiple volumes
  • CHANGE: allow setting READ_ONLY attribute. Fix install and uninstall path. Do not delete storage if not created
  • BUGFIX: (#19054) For Exchange Recovery Pro the creation of the shortcut on RDS fails. Use different sm_client function to obtain the path for the work and the bin directory.

Replication

  • BUGFIX: Do not delete savesets from target store of a replication, if they are active on the target store.

Kernel

  • BUGFIX: (#19254) 'sm_client dir /' (x86) ends with core dump
  • NEW: (#19024) Enable compression of LIS files by default (after 14 days)
  • BUGFIX: (#18711) STPD process does not write more than 64 streams into perf file
  • BUGFIX: (#15290) Suppress with blocking date schedule does only work with first suppress schedule the 2nd suppress schedule is ignored. Loop with index to check all defined suppress schedules
  • BUGFIX: (#19174) Adjust the size of the parent_task for restore_results and migration_results.
  • BUGFIX: (#18914) change the info texts of the data store in case an error concerning capacity and/or hwm occurs.
  • BUGFIX: (#18707) do not set second bit of move medium command since this leads to 'Loader not ready' after some changer movements
  • BUGFIX: (#18958) Function 'close tape': Check for 'action=close' too, as GUI writes only 'close' to DB instead of 'closetape'
  • CHANGE: slu ignores algoritm index 1 when checking encryption capabilities \ IBM drives
  • NEW: Return error message during DB access in 'int_INSERT_INTO_RESULTS()' to caller. At the moment this will always be a 'duplicate key error'
  • NEW: (#18358) Allow more than 2 drives for a Si3 deduplication store
  • BUGFIX: (#19035) Blocking schedules are not blocking anymore. Assure that suppress events are inserted into cal_sheets with unique ids
  • NEW: include the sbc_vss.exe in the GUI and the client installer package.
  • BUGFIX: (#18956) automatically set the vm_server_type to vCenter if the operating system is Windows and the vm_server_type is NULL and the client contains at least one vSphere task.

License

  • BUGFIX: (#18004) vCenter license is calculated incorrectly. Correct scan for 'Datacenter ...:' in output of 'sm_cmd list vSphere'
  • NEW: (#19113) For front side storage license: Count data from successful backups only

Queue Manager

  • BUGFIX: (#18938) Windows: Drive deletion may lead to stalled sm_qm_main process. Avoid to end process group with TerminateJobObject() in iStopProcessGroup() use this function only in JobControl()
  • BUGFIX: (#18938) Windows: Drive deletion may lead to stalled sm_qm_main process. Aborting job process groups in stopped drive queue is done with sending SIGINT, sleep 5s and sending SIGKILL (terminate process) but meanwhile another stop thread may be created to stop the same process group
  • CHANGE: After stopping the service wait until service state is stopped
  • BUGFIX: (#19058) sm_qm sho/full {drive} consumes too much answer time. Do not use asynchron pipe communication for sm_qm_main answer pipe (peek on pipe may cause additional IO and delay)

Installation and Update

  • CHANGE: inserted the new SEP sesam BSR Pro version 10.0.118 in the kit build.
  • BUGFIX: (#19078) When executing set_shortcut, sm_prof.bat is inserted as link in the start menu in the folder SEP sesam\Konfiguration SEP sesam
  • BUGFIX: (#19288) during sesam update do not use vss for backup. If the backup fails continue with the update anyway.
  • BUGFIX: (#19198) Erase typo in MSI installer on service user page

Major fixes and changes v4.4.2.72

For a complete list of GIT change reports, see changes.

GUI

  • bugfix: wrong datastore will be chosen, if datastore table is manually sorted by column
  • BUGFIX: (#18637) In table Tasks->Tasks as List state filter menu offers internal flags instead of description of flags. Duplicates in filter were elimated.
  • BUGFIX: (#18982) To determine an invalid license zip file, the content of the selected zip file is logged if no sm_lic.ini is present in the zip file.
  • BUGFIX: (#18859) force UTF-8 encoding when reading/writing license in API and CLI
  • change: (#18723) Don't show the stack trace scroll pane if no stack trace is shown in the exception dialog
  • BUGFIX: (#18972) Loader action "Export" fails with "drive is invalid", but no drive is needed for the export. Erase the drive check for loader action "Export".
  • NEW: (#17470) Implemented server part for new notification with level confirm but without any GUI popup
  • BUGFIX: (#18893) Start backup as well as create/edit a Schedules->New Backup Event displays wrong Interfaces/Drives
  • BUGFIX: (#18939) Null pointer exception when trying to change comment of restore task
  • BUGFIX: (#18941) Null pointer exception when trying to open a media event; optional field 'label' will not be set
  • BUGFIX: (#18942) Null pointer exception when expanding/collapsing media pool tree node several times at Components->Mediapools.
  • BUGFIX: (#18923) prevent gui from updating results columns that are only written by the kernel
  • BUGFIX: (#18935) Null pointer exception at restore wizard of a VM restore to original
  • change: set textfield 'task' as non editable at tab 'info2' of result dialog
  • BUGFIX: (#18932) Change the calculation of the duration in the backup task is the end time has been changed.

Backup

  • bugfix: (#18965) Line feed in NOT log from vSphere backup. ReadFromHandle() check for signal EAGAIN even if some bytes are already read and add further bytes to buffer
  • BUGFIX: (#18912) Microsoft Exchange: Do not exit sm_reformat_lis over exception handler. The error code 105 is appearing in case of deployed server and client on the one host machine.
  • BUGFIX: (#18963) Microsoft Exchange: sbc reports 'logs are missing' during Exchange full backup. The issue is occurring because of fnmatch() function on Windows (case insensitive) and on Unix (case sensitive).
  • BUGFIX: (#18963) Microsoft Exchange: 'sm_reformat_lis': Return with 'returncode,msg' form all task type specific functions. (real values aren't already set for all of them)
  • bugfix: (#18912) Microsoft Exchange: Do not exit sm_reformat_lis over exception handler. Improved the error handling mechanism.
  • BUGFIX: (#18853) Microsoft Exchange: backup to RDS failed with: ERROR sm_reformat_lis returned exit state 1. Handle the metadata file and check the format, print out more informative error.
  • new: If global variable 'gv_no_metafile_check=1' is set, then exit code from 'sm_reformat_lis' will be ignored
  • BUGFIX: (#18921) Microsoft Exchange: Exchange backups fails with 'ExchangeManagementShell::GetExchangeServer: Value cannot be null' after upgrade from 4.4.1 to 4.4.2. The sm.ini values in section [ENVIRONMENT] must be put in the environment (since version 4.4.2 sm_qm_main is used instead of sm_main)
  • BUGFIX: (#18921) Microsoft Exchange: Exchange backups fails with 'ExchangeManagementShell::GetExchangeServer: Value cannot be null' after upgrade from 4.4.1 to 4.4.2. Add environment variable COMPLUS_ApplicationMigrationRuntimeActivationConfigPath during update not only during installation
  • bugfix: Hyper-V: Operating system was not shown in the not log of a Hyper-V and VMware backup.
  • BUGFIX (#18930) Hyper-V: VSS API error during Hyper-V backup. Handle the Hyper-V 2.0

Restore

  • BUGFIX: (#18955) Restore of system files (e.g. desktop.ini) does not set attributes SYSTEM and HIDDEN. The file creation has to reset the attributes
  • bugfix: (#18976) BSR Pro Windows: Restore from DataStore with slow throughput 30 GB/h - improve FTP data socket creation(fixed first socket connection attempt)
  • BUGFIX: (#18655) BSR Pro Windows: Restore from DataStore with slow throughput 6 GB/h. If DATA socket is closed then stpd 553 message may return WSACONNRESET 10054 or WSACONNABORT 10053 message
  • bugfix: (#18655) BSR Pro Windows: Restore from DataStore with slow throughput 6 GB/h. After reading the data from end of saveset the EOF flag was set but the EOF flag was not reset for further reads

Kernel

  • bugfix: (#16868) Rename 'xxx.GrouWise System 2012' to 'xxx.GroupWise System 20xx'. All occurences renamed
  • BUGFIX: (#18978) BSR Pro Windows: XBSA with HTTP[S] protocol support. Add protocol prefix to PWD file server configuration
  • CHANGE: (#18949) System tray info version is very old. Version set to 4.4.2, Copyright set to 2015
  • BUGFIX: (#18884) Binary files aren't copied correctly over CTRLC (Linux) SMSSH and SSH. Implement binary mode data transfer in PIPE_OPEN_CMD() functions
  • BUGFIX: Correct scan for 'sbc' version string from output of very old Sesam clients. (no 'Build:' tag in version message)

Installation and Update

  • NEW: use the new O&O msi installation package version 10.0.81
  • BUGFIX: (#18947) the virtual file system driver could not be installed on an RDS since the global variables needed were missing. Obtain global variables for pathes via the sm.ini [PATHES] section instead of using the registry values.

Major fixes and changes v4.4.2.70

For a complete list of GIT change reports, see changes.

GUI

  • BUGFIX: (#18922) Avoid NPE at GUI client start
  • BUGFIX: Avoid possible NPE in the new migration dialog
  • BUGFIX: (#18878) Client filter settings are not saved with the view
  • BUGFIX: (#18895) Adding media with defined label was no more possible
  • BUGFIX: (#18864) resetcbt ends with Exception 'Invalid Login to com.vmware.vim25.InvalidLogin', add encrypt vm passwort condition to '>=64'
  • BUGFIX: (#18852) messagebox 'You are about to change the EOL to an already expired point in time' does not appear, when set eol to the past at result dialog, invoked from component 'task results'
  • CHANGE: check, if 'java x64' or 'java x86' is used, when starting VM operation; add log of used 'java version'
  • BUGFIX: (#18903) Duration of schedule isn't read from DB in case of more than 24 hours
  • BUGFIX: (#18873) messagebox 'You are about to change the EOL to an already expired point in time' will not be shown, when set eol to the past at result dialog or if change eol to the past several times and press 'No' several times
  • BUGFIX: (#18873) rewrite term calculator to correctly calculate the next exec time. also add various unit tests to verify correct usage (complete matrix until pcount=4)
  • BUGFIX: (#18873) use Monday as start of week instead of Sunday when working with pcount > 1 in weekly schedules.
  • BUGFIX: (#18873) change existing weekly schedule cause gui freezed, prevent endless loop
  • CHANGE: prevent log stacktrace of 'JsonMappingException', when open backup protocol
  • BUGFIX: (#18886) add schedule via cli does not take duration param -d '6:00'

GUI client (sm_cmd)

  • CHANGE: improve cli restore/restore event help description 'the restore mode of the restore ('o'=overwrite, 'n' no overwrite, 'v'=new version)' for param 'O' (restoreMode)
  • BUGFIX: create selective restore task with cli failed, restoretree='s' will not be set
  • BUGFIX: selective restore with cli failed, restoretype will be set with wrong value

Backup

  • BUGFIX: (#18737) vSphere diff and incr backups hang at "Get changeID" for the second VMDK
  • BUGFIX: (#18881) Starting a task group Hyper-V backup fails. VSS service shouldn't be restarted before the Hyper-V backup process as well as vmms service.

Restore

  • BUGFIX: (#18898) selective restore via cli failed due to wrong created sel file.
  • WORKAROUND: (#....) Restore of migrated saveset with EOM and incorrect label (TAPE99999) in tape trailer: Try to find next label by checking 'DB:result_lbls' from migrated saveset too. Check is done over saveset id from 'DB:restore_results' contains the original saveset name

License

  • BUGFIX: (#18917) VM License has wrong count of various values (datacenter, vm, ...), e.g. all vCenters were counted twice

Installation and Update

  • BUGFIX: (#18896) Changed the default of BSR Pro to leave unchanged instead of install. In addition an installer variable for bsr pro was inserted called BSR_PRO. Additionally the uninstall condition was corrected.
  • BUGFIX: (#18896) Adjust the size of the text in the BSR Pro dialog of the installer so the text does not contain a line break.
  • BUGFIX: (#18892) During update the file name of the original database export file was cut since the size of the variable was not obtained correctly. Use sprintf instead of M_X_SNPRINTF for char*.
  • BUGFIX: (#18797) If a system restart is required after the cbfs driver for the sm_vfs has been installed the msi should now prompt the user for a system restart. The sm_vfs will only work after the system has been restarted. Move the ScheduleReboot custom action to the correct place and insert an additional dialog box explaining the user why the reboot is required.

Major fixes and changes v4.4.2.68

For a complete list of GIT change reports, see changes.

GUI

  • CHANGE: disable button 'back' button at restore wizard, when savesets of selected task will be shown. Chinging a restore task required a reopen of restore wizard
  • BUGFIX: NPE when identifying media pools
  • BUGFIX: Run VM restore cause 'org.dom4j.DocumentException: Error on line 1 of document  : Vorzeitiges Dateiende', when create VM from ovf file
  • BUGFIX: (#18852) In the data store dialog, changing the EOL via the saveset table into a year later than 2036 accidentally changed the EOL to the year 1936.
  • BUGFIX: (#18873) 'Next Execution' time is calculated incorrectly for 2 weeks cycle. Avoid going forward if the target time is today.
  • BUGFIX: In the schedule migration dialog the initialization for the target pool was not correct. Now the target pool is is obtained via the migration task if no target pool has been selected.
  • CHANGE: Improve error message text, when reset CBT via gui failed due to VM snapshots
  • BUGFIX: Reset CBT shows stacktrace, if procedure failed due to still existing snapshots
  • BUGFIX: (#18860) Migration event could not be created or modified. After clicking "OK" to store the migration event, the window "New migration event in schedule..." did not close and the GUI hung. The combo box of the migration event was not filled, when open properties of existing migration event.
  • BUGFIX: (#18869) Any change of the results table in the GUI leads to wrong format of the start time in the results table. This destroys the FDI chain. Leave the start time in the results table unchanged.
  • BUGFIX: VM reset cbt show exception window, if reset CBT failed due to VM snapshots
  • BUGFIX: To ensure that task tree shows all data a Null Pointer Exception check was added
  • BUGFIX: (#18856) Changing device server of a drive deleted the drive. Now the drive is updated instead of deleting it.
  • BUGFIX: device field is disabled and pre-filled for new drive with default type 'LTO'

GUI client (sm_cmd)

  • BUGFIX: (#18850) Allow setting of task group
  • BUGFIX: (#18854) Window 'Current Messages' shows only 'drives'->'system' is missing. Valid line '0>' must be rewritten when message for real drive is set

Backup

  • BUGFIX: (#18853) Exchange Server backup to RDS failed with: ERROR sm_reformat_lis returned exit state 1. Fetch meta file from RDS. Handle the metadata file and check its format, print out more informative error.
  • BUGFIX: (#18834) Migration error - submit/sho_job sm_sbc_com_ext failed. Avoid ERROR_PIPE_BUSY with using semaphore to access qm_server pipe
  • BUGFIX: (#18658) Oracle backup fails with 10054. The backup of an oracle database to an RDS fails to close several part savesets. Add shutdown(SData, SHUT_WR) before calling closesocket(SData) if data is sent to storage
  • BUGFIX: (#18811) Notes backup with linux client fails with XBSA Call BSAEndData (closing saveset). Call DB_Close() before BSAEndData() otherwise the Lotus Notes logasio may keep the DATA socket opened.
  • BUGFIX: (#16206) Validating Backup Integrity By Using CHKSGFILES - enhancement of sm_exchange.py by checking information gathered from NOT file - especially mailbox databases and logs required for bringing one to clean state.

Restore

  • BUGFIX: Mount savesets on RDS running on Windows. Without this fix Exchange Recovery Pro can not be used on an RDS.
  • BUGFIX: (#18835) Restore from path backup fails with: 'Error while unpacking data. Bad archive.'. Change for commit 47135373 for '(#18726) During 'all' backup the same hard linked file is transferred again and again.' did not reset the iRet = MTF_MORE_BLOCKS for the pending directory

License

  • WORKAROUND: Do not count GW user licenses until #18866 (wrong user count for Exchange) is fixed

Installation and Update

  • BUGFIX: (#18849) if the value qm is not present in the sm.ini during the update the service will not be able to start unless the value [DAEMON] 'qm=on' is inserted in the sm.ini. Now the value qm will be inserted in case it is missing.

BSR Pro Windows

  • bugfix: (#18671) BSR Pro Windows: 2nd RETR from Si3 Store failed. Commit 1dfb18d did freed the ThriftClient object but did not reset the openmode
  • bugfix: (#18655) BSR Pro Windows: Restore from DataStore with slow throughput 6 GB/h. STPD: The move operation was only performed for the first data restore request. Now offset is checked again
  • bugfix: (#17807) BSR Pro Windows: O&O DiskImage Backup and Disaster Restore - Call DiskImage via sbc_bsr. A different registry key is used for x86
  • bugfix: (#18529) BSR Pro Windows - Client browser must allow to select BSR Windows for BSR Pro. A different Version key is used under x86: HKLM\Software\Microsoft\Windows\CurrentVersion\Uninstall\{161BB297-50B2-4A3C-8842-A1A00DBE7BA9}