Beefalo V2 4.4.3.86: Difference between revisions

From SEPsesam
No edit summary
No edit summary
Line 83: Line 83:


=== Kernel ===
=== Kernel ===
====Scheduled jobs and external backups fail due to yet uninitilized queue ====
'''''Problem'''''
When a new drive is introduced or a new datastore and media pool are created and backup or some other scheduled job is started immediately after, the job fails with an error about the non-existing queue. The same issue may happen when SEP sesam Server starts after update and backups are started immediately after the  server is started, as ''sm_config_drives'' has not yet created drive queues.
====''sm_sms_watch'' crashes when starting the SEP sesam services or a NEWDAY event====
====''sm_sms_watch'' crashes when starting the SEP sesam services or a NEWDAY event====
'''''Problem'''''
'''''Problem'''''
''sm_sms_watch'' crashes during SEP sesam startup or when a NEWDAY event is started due to the SM_QM status output that leads to overrun in stack during drive configuration.  
''sm_sms_watch'' crashes during SEP sesam startup or when a NEWDAY event is started due to the SM_QM status output that leads to overrun in stack during drive configuration.  
Line 112: Line 115:


=== STPD ===
=== STPD ===
====If a backup is aborted, STPD does not write the metadata (LIS and SGM file), making the saveset invisible and unable to be deleted ====
====If a backup is aborted, STPD does not write the metadata (LIS and SGM file)====
'''''Problem'''''
'''''Problem'''''
''(Applies only to Windows)'' STPD (Sesam Transfer Protocol Server) does not write metadata (LIS and SGM file), causing a saveset to become invisible and unable to be deleted, even though it occupies the space.
''(Applies only to Windows)'' STPD (Sesam Transfer Protocol Server) does not write metadata (LIS and SGM file), causing a saveset to become invisible and unable to be deleted, even though it occupies the space.
====Backup session gets wrong COM answers (not matching the openjob command)====
'''''Problem'''''
Backup session gets wrong COM answers during backup, for example, during Oracle backups using multiple channels or with SAP HANA backups; in the latter case, the backint backup module receives answers that do not match the ''openjob'' command and proceeds without any backup error. Consequently, backups are performed to the arbitrary media pool with a much shorter retention time than configured, which may result in data loss (the backup is not available for restore due to the shorter EOL).


=== SMS ===
=== SMS ===
====Restore of data split on multiple tapes fails due to unsuccessful archive resynchronization====
'''''Problem'''''
Restore of data saved on multiple tapes fails with error: The archive could not be resynchronized. The same correct tape block size is automatically used for the second tape, resulting in failed restore.
====On sayFUSE, EOM is not recognized====
====On sayFUSE, EOM is not recognized====
'''''Problem'''''
'''''Problem'''''

Revision as of 15:36, 18 June 2021

Draft.png WORK IN PROGRESS
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.

Template:Copyright SEP AG en

SEP sesam Beefalo 4.4.3.86 overview

  • <version> released: <date>
SEP Tip.png Tip

New supported systems

No changes from Beefalo V2, see New supported systems from Beefalo V2.

Discontinued systems

No changes from Beefalo V2, see Unsupported OS.

Known issues

Beefalo 4.4.3.86 known issues:

Severity: CRITICAL

Warning sign.pngPossible data loss due to non-executed updates of the entries in the SEP sesam database at extremely high system load.
  • There may be problems updating entries in the SEP sesam database if the system load is so high that the maximum number of connections is reached (PostgreSQL) or the busy timeout (SQLite) is no longer sufficient. This can result in the backup status being incorrectly set as successful and external applications (e.g. SAP Hana) seeing backups as successful even though they did not complete successfully. In such a case, data loss may occur during a restore.
Workaround: SEP is working on this issue and will provide a fix as soon as possible.
Backup jobs fail with MTF analyze errors
  • After updating to 4.4.3.86, several backups end with 553 STOR Failed. MTF analyze error [Error: Invalid MTF descriptor block checksum. (53)].
Workaround: SEP is working on this issue and will provide a fix as soon as possible.
Backup to tape succeeds despite IO error, but restore fails
  • An IO error can occur during a backup, for example, if the tape is forcibly unloaded from the drive. A backup is still successful, splitting the saveset between two tapes. However, the restore fails with Error: Cannot read from remote archive.
Workaround: SEP is working on this issue and will provide a fix as soon as possible.
SEP sesam v. 4.4.3.84 – (Linux) Backup to VTL library* completed successfully although media write errors were reported and the backup could not be completed
  • (Applies only to Linux) When backing up to VTL, the VTL library cannot continue writing to the underlying block device due to insufficient space and tracks error. SMS ignores the error and continues writing the data to VTL without sufficient space. Backup log shows backup completed successfully although it actually failed and no data was written to backup media.
    (*This behavior is seen with the QUADStor VTL only.)
Workaround: Download the hotfix from SEP Download Center to your Linux server or RDS and install it by executing the following commands as root in a terminal session. Attention: This operation will stop all running backups.
#>sm_main stop node
#>cd /opt/sesam
#>tar xvzf <hotfix file> 
#>sm_main start node

Severity: MINOR

SEP sesam v. 4.4.3.84 SP2 – Automatic SP update of Windows clients on the Linux Server fails
  • When trying to update the Windows clients of the SEP sesam Linux Server, automatic SP update fails or is stuck in a loop due to the missing sm_execute_update.cmd file.
Workaround: Download the sm_execute_update.cmd file manually from the Linux service pack directory and copy it to the skel directory of the SEP sesam Linux Server, for example, sesam:/opt/sesam/skel # ls -l sm_execute_update.cmd.
Then start the SP2 update of the SEP sesam Linux Server via GUI.
SEP sesam v. 4.4.3.84 SP2 – After SP2 is installed, additional Python lib file for Debian Stretch is required
  • After SP2 is installed, there is an issue with the Python lib file for Debian Stretch on the SEP sesam Server. For example, backup savesets are no longer purged (deleted) from data stores.
Workaround: After the SP 2 installation is complete, you must download and manually copy the Python lib file for Debian Stretch to /opt/sesam/bin/sesam on SEP sesam Server: download.sep.de/servicepacks/4.4.3/4.4.3.84/linux/stretch/libpython2.7.so.1.0
SEP sesam v. 4.4.3.84 – Dell EMC Unity NDMP client configuration issue
  • (Applies only to Dell EMC Unity NDMP backup client) When configuring a Dell EMC Unity NDMP client in the SEP sesam GUI, it is not possible to select DELL EMC Unity as the operating system.
Workaround (SP2 installation is not required): When configuring a new Dell EMC Unity NDMP client, select EMC VNX as the operating system; for more details, see Adding NDMP host as a SEP sesam client. SEP is working on this issue and will provide a solution as soon as possible.
SEP sesam v. 4.4.3.84 – In the backup task for Dell EMC Unity NDMP client, the wrong source is shown
  • (Applies only to Dell EMC Unity NDMP backup) When creating a Dell EMC Unity NDMP backup task in the GUI and browsing for a source, NetApp is incorrectly shown in the Client File View and must be ignored.
Workaround (SP2 installation is not required): Make sure to select the appropriate NDMP source for Dell EMC Unity NDMP backup. SEP is working on this issue and will provide a solution as soon as possible.
SEP sesam v. 4.4.3.84 SP1 & SP2 – The update status is not updated after manual update installation
  • After manual update of one or several SEP sesam Clients, the Topology view shows incorrect update status. This status cannot be reset and it will continue to display a yellow arrow on the client even though the client is already updated or if the update has failed.
Workaround: As the update-related fields in the client's table are not automatically cleared after the manual update, execute the following command for each affected client and replace the CLIENTNAME variable with the actual name of the client in SEP sesam.
sm_db "update clients set available_version='',available_number=NULL,servicepack_available_number=NULL,jar_available_version=NULL,jar_available_number=NULL where name='CLIENTNAME'"
SEP is working on this issue and will provide a solution as soon as possible.
SEP sesam v. ≥ 4.4.3.82 – (Windows) SEP sesam GUI cannot be opened when the group policy option Prevent access to registry editing tools is applied
  • (Applies only to Windows) Running the GUI fails after updating SEP sesam to version ≥ 4.4.3.82. This issue seems to be related to the group policy (GPO) setting: if Prevent access to registry editing tools is applied for a specific user or group, opening the GUI does not work for the users which are prevented from accessing the Registry. In most cases, users with admin privileges will still be able to run the GUI as they are typically exempt for a GPO (and are able to access the Registry). If this policy is disabled (or not configured), the users may access the GUI regardless of their privileges.
Workaround: To enable access to the GUI for the affected users (typically non-admin users), disable the group policy option Prevent access to registry editing tools. Note that only experts should edit the registry; modifying Windows Registry incorrectly might crash your Windows operating system causing data loss.
  1. Press the Windows key + R to open the Run window.
  2. Type gpedit.msc and press Enter.
  3. In the Group Policy Editor -> User Configuration -> Administrative Templates -> System, double-click on Prevent access to registry editing tools setting in the right panel and select Disable. Then click Apply.

Enhancements and changes Beefalo 4.4.3.86

Consistent export of the SEP sesam Server database

SEP sesam used to export its database table by table without taking care of foreign keys. As foreign keys enforce the database referential integrity and ensure that the relationship between database tables is preserved during the insertion of data, the export is now improved by using pg_dump, resulting in a consistent export and a faster import of the database.

Optimized license check

License check performance has been improved; the license is now checked during start-up, reducing the noise and cluttering the log with detailed logging on license checks.

Fixed issues Beefalo 4.4.3.86

Kernel

Scheduled jobs and external backups fail due to yet uninitilized queue

Problem When a new drive is introduced or a new datastore and media pool are created and backup or some other scheduled job is started immediately after, the job fails with an error about the non-existing queue. The same issue may happen when SEP sesam Server starts after update and backups are started immediately after the server is started, as sm_config_drives has not yet created drive queues.

sm_sms_watch crashes when starting the SEP sesam services or a NEWDAY event

Problem sm_sms_watch crashes during SEP sesam startup or when a NEWDAY event is started due to the SM_QM status output that leads to overrun in stack during drive configuration.

Sesam DB

If the connection limit to the PostgreSQL database is reached, SEP sesam Server stops running

Problem By default, the Sesam DB instance runs with the default max_connections = 100. When this connection limit is reached, for example, to the PostgreSQL database, SEP sesam terminates but the error log does not contain information on database access errors and reaching the maximum connection limit.

SEP sesam v. 4.4.3.xx – Check status of 'sbc_com closejob' call before backup is reported as successful

Problem SEP sesam external backup modules (Oracle, Hana...) report backups as successful after the data transfer is complete. Any problems during the 'closejob' operation are ignored. If, for example, the copy of the SGM file fails, the saveset will not be found during restore. Checksum errors may also be detected, but the backup is still shown as successful.

Update

SEP sesam v. 4.4.3.85 – Linux update fails due to faulty copy DB file command

Problem On some operating systems (e.g., SLES12) the behavior of overlapping string operations is undefined; because the update script uses the same string as source and target in the string operation, the update fails.

Windows client update from GUI fails if BSR is already installed on the system

Problem If BSR is already installed on the system to be updated automatically from GUI, the update fails due to a syntax error in the cmd file.

Installing SEP sesam BSR Pro fails due to the wrong installation path

Problem Installing SEP sesam BSR Pro fails due to the wrong installation path with incorrect double slash in the BSR Pro path.

STPD

If a backup is aborted, STPD does not write the metadata (LIS and SGM file)

Problem (Applies only to Windows) STPD (Sesam Transfer Protocol Server) does not write metadata (LIS and SGM file), causing a saveset to become invisible and unable to be deleted, even though it occupies the space.

Backup session gets wrong COM answers (not matching the openjob command)

Problem Backup session gets wrong COM answers during backup, for example, during Oracle backups using multiple channels or with SAP HANA backups; in the latter case, the backint backup module receives answers that do not match the openjob command and proceeds without any backup error. Consequently, backups are performed to the arbitrary media pool with a much shorter retention time than configured, which may result in data loss (the backup is not available for restore due to the shorter EOL).


SMS

Restore of data split on multiple tapes fails due to unsuccessful archive resynchronization

Problem Restore of data saved on multiple tapes fails with error: The archive could not be resynchronized. The same correct tape block size is automatically used for the second tape, resulting in failed restore.

On sayFUSE, EOM is not recognized

Problem The EOM (End of Media) (EOM) parameter is not recognized on sayFUSE, causing backups to fail for seemingly no reason.

Next EOM after EOM IO error sets IO error flag

Problem SMS remembers EOM IO error status on subsequent normal EOM handling and incorrectly sets IO error flag on subsequent EOM.

Backup to tape during IO error is successful but restore fails

Problem Backup to tape during IO error is successful, but restore fails with Error: Cannot read from remote archive.

Media

Wrong start time is displayed for readability check

Problem Start time for media readability check is set in the past.

Virtual environment

Restored OpenNebula VM does not boot

Problem Restored OpenNebula virtual machine fails to boot because the restored image was created with the wrong format setting.

OpenNebula VM architecture is not restored

Problem For OpenNebula virtual machines, the architecture (i386, x86_64) can be specified when the VM is created. During restore, this option is not restored according to the specified VM xml configuration.

DB modules

Saveset inquire for external backups (Oracle, SAP...) returns failed backups

Problem When searching for savesets related to external backups, savesets with failed status are returned.

Restore

Restore exclude filter does not work correctly

Problem The restore exclude filter does not work correctly. For example, for a selective restore, the exclude filter is set to "-*.ini" (- prefix), but the filename is not extracted correctly (e.g., /debug.ini instead of debug.ini) thus the matching file cannot be found and excluded.

Restore from migrated saveset uses wrong transport mode

Problem Restoring a migrated backup shows settings of the original backup in the restore wizard with the selected transport mode, then uses the correct RDS without adapting the protocol of the given interface (transport mode), for example, instead of using HTTP the FTP mode is used, which leads to failed restore.

Migration

Core dump occurs after starting the migration

Problem Migration is terminated (core dump of sm_sms_copy) with exception 0xc0000417 after a NEWDAY event occurred.

Migration is succesful with dabase error

Problem Migration results in database error DB_SEL_GETVALUE: Error during DBaccess - return 0. Incorrect szSql variable was used to execute DB select.

Core dump occurs after the migration was started

Problem The migration is terminated (core dump of sm_sms_copy) with exception 0xc0000417 after a NEWDAY event has occured.

NDMP

NDMP restore from tape fails

Problem NDMP restore from tape failed with NDMP: 1 info: LOG_MESSAGE: 'RESTORE: Mangled directory' due to the incorrect buffer handling with new asynchronous read approach.

SBC Linux

Restore exclude filter does not work correctly

Problem The restore exclude filter does not work correctly. For example, for a selective restore the exclude filter is set as "-*.ini" (- prefix), but the filename is not extracted correctly (e.g., /debug.ini instead of debug.ini) thus the matching file cannot be found and excluded. Another example of the erroneous exclude filtering is failing to reset after the first match so all the following items are excluded, too.

SBC Windows

=Windows backup of volume mounted as folder name fails

Problem On Windows, backup of a volume that is mounted as a folder name (instead as a drive letter) fails with Invalid file name in FILE block.