Change report Jaglion V2 Service Packs

From SEPsesam
This article is in the initial stage and may be updated, replaced or deleted at any time. It is inappropriate to use this document as reference material as it is a work in progress and should be treated as such.

Change Report for Jaglion V2 SP1, SP2 and SP3

Updated: March 6, 2023 | The purpose of this change report is to list the changes, enhancements and fixes in the SEP sesam software between versions Jaglion V2, Service Packs, Jaglion V2 SP1 and Jaglion V2 SP1. Jaglion V2 SP1

Change report
  • BUGFIX: Initialization of an already used tape media can take longer than one hour. Solution: Optimize SQL statements in 'REMOVE_SAVESET()' function
  • BUGFIX: Multistream backup with EOM fails with "Failed to write Tape Trailer. Tape is already closed.". Problem: Race condition while different backups try to write Tape Trailer. Solution: Tape Trailer write should be done under mutex
  • BUGFIX: Tape init must ensure to remove meta files of older savesets as well. The old saveset counter in the file name must be set accordingly.
  • BUGFIX: Command event cannot start a shell script, e.g.
  • BUGFIX: Command event doesn't start if penultimate parameter in command line is '-d'
  • CHANGE: Select statement in 'sm_remove_saveset purge' takes 25 minutes to finish. Add index for DB:restore_results.saveset_used
  • BUGFIX: Saveset delete fails with 'Error: Option -d or -L is required'. Use correct double quotes for '-e' paramter so it works on Windows correctly
  • BUGFIX: Hyper-V restore from RCT backup fails to add VM to cluster. Added missed function "AddVmToCluster(...)" call.
  • BUGFIX: Hyper-V restore of running VM fails with wrong error message "Failed to call "DestroySystem" method". Added check for != "Not Running".
  • BUGFIX: Sesam Master GUI throws exception when adding a new remote server
  • BUGFIX: Fix NPE in Java restore wizard when starting existing restore tasks Jaglion V2

Change report
  • BUGFIX: End time from start time frame in DB:cal_sheets is not adjusted from schedule
  • BUGFIX: Can't delete task event in CLI
  • BUGFIX: Can't delete command event in CLI
  • BUGFIX: Can't define loader ctrl type for loader in CLI
  • BUGFIX: Unclear error printed when parameter for 'resetcbt' command contains invalid values
  • BUGFIX: Java exception during creation of more than 2 Si3-NG data stores
  • BUGFIX: Relocation at MSSQL shows just one entry due to wrong format at catalog file
  • BUGFIX: Restore fails because sometimes the wrong segment number is used
  • BUGFIX: Aborting a MS Exchange backup sets wrong state, so GUI does not show it anymore
  • BUGFIX: MS Exchange backup finishes with warning when VSS snapshot cannot grow anymore
  • BUGFIX: Primary key error in DB:cal_sheets, because same id created for migrations and restores
  • BUGFIX: sm_reformat_lis fails with 'invalid start byte' when MSSQL manual backup contains mutated vowels
  • BUGFIX: Manually removing saveset from the original store after migration removes backup's results entry
  • BUGFIX: Fixed VMware network mount for a VM backed up on Windows Sesam server or Windows RDS, mounted on a Linux client.
  • BUGFIX: Write DB:migration_results.target_saveset for replication tasks too, so further migrations will find DB:migration_results entry
  • BUGFIX: SIR doesn't work if 'guestmount' will be started by 'sm_sshd'. Start 'guestmount' in new process group
  • BUGFIX: Fixed RHEV mount for RHEV backups
  • CHANGE: Improved Si3NG
  • CHANGE: Powerscale backups save less data than with Version 4.4.3. Problem: Folder '$WINDOWS.~BT' must not be excluded by hard coded pattern.
  • BUGFIX: Next Execution time is calculated incorrectly because of the summer-winter-time change
  • BUGFIX: Restore of Nutanix with VM with 2 disks fails
  • BUGFIX: VFS mount of a Nutanix VM Incremental backup
  • BUGFIX sm_qm_main log gets too large due to loop with shared drives configuration
  • BUGFIX: Full support for Hyper-V RCT COPY/FULL/DIFF/INCR with VFS mount (SIR)
  • BUGFIX: Enhance openjob/closejob handling of external jobs.
  • BUGFIX: closejob of external requests takes 10 minutes when no saveset has been specified
  • BUGFIX: All volumes are backed up, if source have name with "all" at the end (e.g. E:/AG,E:/ALL).
  • CHANGE: Show the total disk space required by VM (VHDX+AVHDX) not only the VHDX during VM browsing in GUI
  • CHANGE: Hyper-V VM backup - Added the consistency level switching possibility (application or crash)
  • BUGFIX: Switch to FULL backup for Hyper-V RCT if a new volume was added between FULL and INCR.
  • BUGFIX: sm_dedup_interface is not able to delete savesets in specific case when switching to a 'plain' DS.
  • BUGFIX: Path generation restore of Diff-Full chain of migrated UNC backup exits with 'Error while unpacking data. Bad archive.'
  • CHANGE: Set correct Java parameters to start GUI client on Linux using Java 17
  • BUGFIX: Selective restore of migrated saveset fails because SEL file with restore ID of backup could not be copied to RDS
  • BUGFIX: Deadlock occurs when CHECK_SPARE cannot retrieve SPARE_ tapes
  • BUGFIX: The restore of a .nvram works only with the same folder name
  • BUGFIX: Client report cannot be opened from Java UI via context menu of 'Topology' view.
  • BUGFIX: A new super user can be created via "sm_cmd add user ..."
  • BUGFIX: Can't define loader ctrl type for loader in CLI
  • BUGFIX: Can't delete taskevent via the CLI
  • BUGFIX: Can't delete command event via the CLI
  • BUGFIX: End time from start time frame in DB:cal_sheets is not adjusted from schedule
  • BUGFIX: 'Manage VM tasks' generates wrong CLI command according vSphere Tags or custom attributes
  • BUGFIX: Specific vSphere VM is not shown when using filter option
  • BUGFIX: vSphere virtual machines with specific special characters are not shown
  • CHANGE: Disable media pool crypt key length constraint for release branch as customers used longer keys than allowed with SQLite
  • BUGFIX: VDDK disconnect is not executed as separate call when sbc cores
  • BUGFIX: CLI does not show content for some values of a schedule
New features
  • NEW: Termination of SEP sesam command events, wildcards '$PID' and '$CMD' can be used in abort command
  • NEW: Extend sbc -k function to send signal to non sbc programs
  • NEW: GUI - Termination of Sesam Command Events
  • NEW: Backup with many sources should not fail if one of the sources does not exist. Use option '-o ignore_missing_source'
  • NEW: Exclude '.vmem' files during vSphere backup
  • NEW: Nutanix VHD/VFS mount for Windows
  • NEW: Java wrapper for all Sesam components. Read and set 'sm.ini [Java]->sm_java_options' for all calls in 'sm_java'
  • NEW: Allow 'postgresql-14' for Sesam server installation
  • NEW: Use 'postgresql-14' for Sesam server installation on Ubuntu 20.04
  • NEW: GUI - Allow pre authenticated users by external authentication source (X-SEP-Authenticated). Jaglion V2 SP1

Change report


  • BUGFIX: Web-UI - VMware "Restore to original VM" not possible
  • BUGFIX: Web-UI - Selection of a new Hyper-V restore target is ignored
  • CHANGE: Web-UI - Correct links to Release Notes Jaglion V2
  • BUGFIX: Do not start backup with empty saveset name, DB2 backups have been started without saveset name
  • BUGFIX: RHEV: Catch Index Error during restore to new VM
  • BUGFIX: Canceling vSphere backups does not work correctly if started by a task group. Add 'order by start_time' to SQL statement to get the correct sub-task entry from 'DB:results'
  • BUGFIX: Analyze of backup log with warnings leads to failed backup status: 'SBC logging without final state message'. Set at least one warning message in analyze backup log
  • BUGFIX: 'sm_sshd' executes 3 'icacls' processes during each processs start.
  • BUGFIX: Queued job 'sm_sbc_com_ext' is not removed when SAP Hana backup runs into timeout
  • BUGFIX: Notification about corrupted DataStores is no longer sent.
  • BUGFIX: Windows: sbc crashes on path backups with large exclude lists.
  • BUGFIX: Restore of UNC folder runs into error because parent folder cannot be overwritten
  • BUGFIX: sm_data_store check_db -o remove all deletes all files from datastore if a file named '.data' found on DataStore Jaglion V2

Change report
  • BUGFIX: Since Jaglion V2 the configured clients in command events have been ignored. Now they are used again, as default when creating a new command event or in case of immediate start.
  • BUGFIX: In WebUI the complete restore of VM did not create 'nvram' by default
  • NEW: Allow to disable the HTTP protocol use of the SEP sesam REST server
  • BUGFIX: Selective Path restore - not all required items are restored. Problem: Journal file (LIS) with different encoded lines (ISO8859 and UTF-8)
  • BUGFIX: Selective restore from SAYTEC sayFUSE devices failed with 'SMS user is not allowed to save on this directory'
  • BUGFIX: Linux path backup ends with warning (List of EA returned with error: errno = 34, error = Numerical result out of range). Increased buffers.
  • BUGFIX: Linux backup failed with 'Call to sbc_regcomp_all() failed' when exclude contains a comma within brackets. Solution: If comma prefixed with slash then its treated as regex symbol and not as separator.
  • BUGFIX: vSphere SFR failed with object is not subscriptable using WebUI, if DIFF saveset is on a different pool
  • BUGFIX: vSphere SFR failed when migrated backups from tape have been migrated back to DataStore
  • BUGFIX: XenServer VM restore with umlauts in its VM-disk description
  • BUGFIX: Si3 maxPage calculation of NG must be done correctly for bigger stores
  • BUGFIX: Since update to on Linux tape encryption could not be turned on during backup. Solution: Due to a variable overflow the salt key was not set correctly
  • CHANGE: Si3-NG does not work on a NSS volumes. Add file system type 'nssvol' to Si3 FS blacklist, so server will not start on it
  • BUGFIX: All MS SQL data bases are backed up, if source db name starts with 'all' (e.g. db name is 'allris')
  • BUGFIX: HCL Domino incremental backup 'The number of objects saved from SBC (...) and the (0) LIS file does not match'
  • BUGFIX: Backup of source with special characters does not work (since version 5.0) Jaglion V2

Version contains a bugfix for deletion of manually added exclude patterns that happens during update to version Jaglion V2 on backup clients. Version includes all improvements and bug fixes of SP3. It is not required to use lower versions and service packs and they may even remove exclude patterns. Update from Beefalo V2 (e.g. or .86) should be done with Jaglion V2. Jaglion V2 SP3

Service Pack 3 solves the purge process problem (introduced in v. and contains some other fixes and changes.

Change report


  • This SP3 fixes the problem with the purge process that was reintroduced in version and could lead to the accidental removal of an otherwise valid backup.
  • Windows: sm_qm hangs if Sesam Queue Manager (sm_qm_main) is terminated.
  • Linux SBC skips the first directory if the backup source directory contains a .nosbc file.
  • After applying Service Pack 1 to version, on Linux exclude list processing via the exclude file is ignored.
  • The data store file system check (FSCK) does not take care of ongoing migrations.
  • Backing up to a media pool with no free tapes will not work, even if it’s allowed to use media from other pool.
  • Sesam Queue Manager allocates too much memory. When configuring a shared drive, there were problems with drive synchronization.
  • The INC/DIFF saveset also increases the EOL of other differential or incremental savesets (other DIFF/INCR chain).
  • Incremental SEP sesam backup saves too much data after starting SEP sesam.
  • Selective restore may show files with timestamps from newer backups than originally selected.
  • Microsoft SQL Server INC/DIFF backup fails with "Mismatch database_backup_lsn [...] full first_lsn".
  • Savesets created by a failed migration must not be used for restores or further migrations.
  • Migration from tape copies the same saveset multiple times if EOM occurred during backup to tape.
  • Hyper-V backup results in sm_reformat_lis-error "'utf-8' codec can't decode byte".
  • User in command event is ignored.
  • Adapt restore mount option if restore task is restarted.
  • Restore using 'mount' fails, if migrated saveset is involved.
  • Client update via GUI hangs in "progress" state.
  • Do not purge savesets used for restore.
  • Failed Microsoft SQL Server incremental backup must switch to FULL.
  • vSphere SFR via WebUI might fail with 'Error during remote copy' if a migrated saveset is involved.
  • The "Release Notes" button in the 'License Agreement' window of Jaglion V2 leads to the previous release notes.
  • Support for OES 2022
  • "Checksum of saveset copy does not match" when multiple entries with the same backup_id exist in result_lbls
  • Restore of Citrix XenServer v4_4_3 backups stored on RDX media fails
  • Manually added exclude patterns have been deleted during update
  • Installation of SP on Windows fails if the installation directory is not C:

New Features

  • Support for Open Enterprise Server 2022
  • Mount Nutanix VMs on Windows Sesam Server Jaglion V2 SP2

  • SP2 installation is highly recommended as it includes various performance, stability, and other key customer fixes not available in the initial release of Jaglion V2 or SP1.
Change report
  • Since SEP sesam SP1, the old syntax for regexp excludes no longer works for directories. Due to this exclude issue, Linux file system backups may store more data than expected. Note: All customers who have adjusted their exclude patterns after installing SP1 must now:
    • either add the following switch to the backup options in the GUI to distinguish between files and folders using ending "/$": -o excl=regexp_split
    • or reverse their patterns for directories by removing the "/$" suffix!
  • Migration does not compare checksum anymore since Jaglion.
  • sm_reformat_lis cannot copy the SharePoint Server LIS file. Jaglion V2 SP1

  • You must install Jaglion V2 before you install SP1.
Change report
  • BUGFIX: (#30912) Concurrent backups result in backing up incorrect Nutanix VM snapshot disks because the same UUID was incorrectly used to create the VM snapshot. Solution: A unique client identifier has been added for querying the Nutanix UUID. If the UUID of the backed up VM and the UUID of the snapshot information do not match, the backup terminates with an error.
  • BUGFIX: Fixing the double free problem in qm_main that can trigger qm_main crash
  • BUGFIX: (#30895) External job stays in state queued, if submit timeout expires. Remove wrong check for backup operation.
  • BUGFIX: (#30745) Sesam bootstrap DB export doesn't contain information about migrated Sesam backup. Correct SQL query to use 'backup_id' instead of 'saveset'
  • BUGFIX: (#30784) 'Cancel Current Running Activities' updates after 100 seconds the state of all still running tasks. Solution: Do not update results' state
  • BUGFIX: (#30868) 'DB:results.stop_time' written in wrong format. Initialize 'szStopTim' with empty string, so 'time_get()' will return correct time in correct format
  • BUGFIX: (#30822) Microsoft Windows Unicode with 4 Bytes could not be converted on Linux RDS. Solution: Use iconv() to convert from UTF-16 to UTF-8
  • BUGFIX: (#30822) Microsoft Windows Unicode with 4 Bytes could not be converted on Linux RDS. Solution: Use iconv() to convert from UTF-16 to UTF-8
  • NEW: (#25577) Use HTTP(S) to cancel STPD of running tasks. If 'gv_conf_use_com_stpd_kill=http/https' set http or https connection to STPD will be used to terminate active backup/restore connections
  • NEW: (#30781) Add LTO-9 drive and media type
  • BUGFIX: (#30496) Backup of ACL on CIFS share failed with ERROR_PRIVILEGE_NOT_HELD (1314) 'A required privilege is not held by the client'. Solution: Use CreateFile() File Security and Access Rights according to
  • BUGFIX: (#30790) sm_qm_main. When changing que back to idle all relevant queues that share a share should be set to idle change: modified qm_main unit tests to use unit test framework change: modified sm_qm_main unit tests to reflect latest changes BUGFIX: (#30550) - pending jobs never got processed or got processed very late. Problem caused by incorrect priority check
  • BUGFIX: (#30669) License check sends notification for end of maintenance instead of license expiration
  • BUGFIX: (#30757) sm_break don't kill the parent restore task. Problem: Process was terminated but state was still active. Solution: In no process found then update database
  • BUGFIX: (#30764) Exclude deactivated VM tasks from the license count
  • BUGFIX: (#30772) Status of data store remains OK if RDS is not reachable and 'disk stat only' flag set
  • BUGFIX: (#30761) Fixed generation SFR of vSphere VM of backup subtype Image. sm_vmdk_interface had a condition to skip adding of FDI chain savesets to LSL file for VFS in prepare_vm_vmdks(). Old code that was not properly reworked within changes for common sm_vfs.
  • NEW: (#30861) Check, if all vSphere configuration files are saved, which are mandatory for restore
  • CHANGE: New version for release build
  • CHANGE: (#30792) Saveset will not be removed from data store, if immediate delete is selected in GUI. Add 'purge option for 'sm_sesam remove_saveset -s ssid -o purge...' call, to execute removal of saveset file from DS
  • BUGFIX: (#30860) Exception with 'syntax error at or near "AND"' occurs if three backup filters are active at 'Monitoring/Last Backup State'
  • BUGFIX: (#30862) Type filters 'Migration' and 'Replication' at 'Monitoring/Last Backup State' have no function since version 5.0.0
  • BUGFIX: (#18079) Backup of virtual machines not possible if virtual machine data resides in subdirectory on data-store
  • CHANGE: (#30865) Try to import OVF file during VM restore with original network setting
  • NEW: (#30221) Remove support for Java 1.8
  • NEW: (#30806) Allow separate exclude for files and directories. If exclude ends with "/", it should be seen as directory regex, if no "/" at the end, is seen as file regex