Known Issues and Limitations: Version 4.1 IBM Spectrum Protect™ Snapshot - all Unix Components

Content

Tip: Beginning with Version 4.1.3, IBM Tivoli Storage FlashCopy® Manager is now IBM Spectrum Protect™ Snapshot. Some applications such as the software fulfillment systems and IBM License Metric Tool use the new product name. However, the software and its product documentation continue to use the Tivoli Storage Manager product name. To learn more about the rebranding transition, see http://www.ibm.com/support/docview.wss?uid=swg21963634.


The following IBM Spectrum Protect™ Snapshot components share the same known issues and limitations. Therefore, the known issues and limitations for all of the following components are grouped together in a single document. The components are:

• IBM Spectrum Protect Snapshot for DB2
• IBM Spectrum Protect Snapshot for Oracle
• IBM Spectrum Protect Snapshot for Oracle (in SAP® environments)
• IBM Spectrum Protect Snapshot for Custom Applications

• Applicable to All IBM Spectrum Protect Snapshot Unix Components
• Specific to the IBM Spectrum Protect Snapshot for DB2 Component
• Specific to the IBM Spectrum Protect Snapshot for Oracle (in SAP® environments) Component
• Specific to the IBM Spectrum Protect Snapshot for Oracle Component
• Specific to the IBM Spectrum Protect Snapshot for Custom Applications Component
• Specific to the AIX Platform
• Specific to the Linux Platform
• Specific to the HP-UX Platform
• Specific to the Solaris Platform
• Specific to Virtualized Environments
• Specific to IBM XIV® Storage System hardware
• Specific to IBM System Storage SAN Volume Controller (SVC), IBM Storwize® family, IBM Flex System™ v7000, and IBM FlashSystem™ V840
• Specific to IBM System Storage DS8000 hardware
• Specific to IBM Spectrum Scale (formerly IBM General Parallel File System (GPFS™))

• Resolved Issues and Limitations Specific to IBM XIV® Storage System hardware with IBM Spectrum Protect Snapshot 4.1.1

• Resolved Issues and Limitations Specific to the IBM Spectrum Protect Snapshot for Oracle Component with IBM Spectrum Protect Snapshot 4.1.3

• Resolved Issues and Limitations Specific to the IBM Spectrum Protect Snapshot for Oracle Component with IBM Spectrum Protect Snapshot 4.1.4

• Updates, clarifications, and corrections to the currently published version of the IBM Spectrum Protect Snapshot V4.1 publications are located in the Documentation Updates tech note at: http://www.ibm.com/support/docview.wss?uid=swg27039330.

• Insufficient space on root file system
  
  If the root file system is running out of space during IBM Spectrum Protect Snapshot installation and/or setup, the process might hang. Always ensure there are sufficient space reserves in the root file system.

• Installation in console mode requires X Window server
  
  In combination with some ssh clients, the console-based installation incorrectly checks for the presence of a X Window server. Therefore, if a X Window server is not available during IBM Spectrum Protect Snapshot installation you may see this error:
  
  Launching installer...
  
  X connection to localhost:10.0 broken (explicit kill or server shutdown).
  
  User response: Unset the DISPLAY environment variable (unset DISPLAY) and re-run the installer. (77219)

• When uninstalling in console mode , the uninstaller may report "Some items could not be removed" although uninstallation completed successfully. (77924)

• "Previous" button of uninstallation agent
  
  Using the "Previous" and "Next" buttons in the uninstallation wizard after the package has already been uninstalled may lead to an exception. (78014)

• Concurrent fcmcli operations on the same backup server
  
  An operation "fcmcli -f tape_backup" (e.g. off-loaded backup to TSM) running on the IBM Spectrum Protect Snapshot Backup Server may terminate with the message:
  
  FMM6276E Another mount agent service is currently running
  
  when another fcmcli operation is running a second operation on the same IBM Spectrum Protect Snapshot Backup Server.
  
  Work-around: Use the profile parameter NUMBER_BACKUPS_IN_PARALLEL in the OFFLOAD section of the profile to specify the number of DB2 partitions to be backed up in parallel per IBM Spectrum Protect Snapshot backup server. If two or more DB2 partitions are backed up by the same IBM Spectrum Protect Snapshot backup server you can prevent parallel operations by choosing this number appropriately.

• Logging in the case of off-load operations
  
  When a customer starts multiple IBM Spectrum Protect Snapshot off-load operations (mount, unmount or tape_backup) in parallel, then the IBM Spectrum Protect Snapshot detailed log output gets written to the log file of the command that was started last. Also, some early log messages of a subsequent mount or unmount operation may be written to the detailed log of a previous off-load operation.
  No detailed log output is lost, but the log output may not be found in the detailed log file where it would be expected.
  
  Work-around: If detailed log output is missing for a specific off-load operation check the detailed log files of all IBM Spectrum Protect Snapshot operations that were executed in parallel. (84855)
  Early messages of mount or unmount operations may be found in the detailed log of a preceding off-load operation.

• Backup and delete operations running concurrently
  
  If a FlashCopy backup operation is started while a delete operation with the force option is still running the backup operation may fail with message FMM4008E logged to the summary log file:
  
  DB 00:25:50 (712) FMM4008E File '/db2/S43/sqllib/acs/acsrepos/ori.1313619101000' cannot be opened. Reason: errno(2) No such file or directory
  DB 00:25:50 FMM9223E The operation will be aborted.
  
  Work-around: Make sure the delete operation completed by monitoring the log file for FlashCopy withdraw messages, then re-start the backup operation.
  Please note: A delete operation may take several minutes to complete.
  (71973)

• Concurrent mount commands
  
  Running multiple mount commands concurrently on the same machine for the same database instance is not supported. Attempts to do so may cause all requests to fail.

• Inquire command and background reconciliation interference
  
  If an fcmcli -f inquire command interferes with a background reconcile operation (performed by default every 12 hours), the fcmcli -f inquire process might fail and print errors about snapshots that are not valid anymore. For example:
  
  ***********************************************************************
  db2acsutil load /db2/R95/sqllib/acs/libacsdb2.a options @/db2/R95/tc_tool/tc/Temp/db2acs.cfg query status show details
  **********************************************************************
  
  ERROR: Error initializing query statement. (rc = 18)
  db2ACS_ReturnCode:
  returnCode: 18
  reasonCode: 0
  description: FMM8323E IBM Tivoli Storage FlashCopy(R) Manager : Problem occurred while
  processing db2ACSBeginQuery().
  Please check log file:
  acsprod1:/db2/R95/sqllib/acs/logs/summary.20100422.log,
  /db2/R95/sqllib/acs/logs/details/inquire.20100422135640,
  /db2/R95/sqllib/acs/logs/summary.20100422.log for more information.
  
  Error information: 'FMM4008E File '/db2/R95/tc_tool/tc/Temp/acsrepos/ori.1271936982492'
  cannot be opened. Reason: errno(2) No such file or directory
  
  summary.log:
  ====================================================
  New query operation started for database instance *, database *.
  ====================================================
  DB 13:56:40 (66e) FMM1510I New connection received.
  DB 13:56:40 (66e) FMM1513I *****> Database client connected: instance *, database *, partition NODE0000
  DB 13:56:40 (66e) FMM1515I Client is logging to acsprod1:
  MON 13:56:40 (507) FMM0104I Deleting the data container '
  string=8,NODE0000
  string=9,R95log0vg
  string=10,A0G8BI7A29' from the backup 'db2r95,R95,20100422134304' in the repository.
  IBM Tivoli Storage FlashCopy(R) Manager found that the volumes of this data container do not contain a valid snapshot in the storage system.
  
  Those errors can be ignored. They will not be displayed any more when the "inquire" command is run again.

• Using csh may cause cloning operations to fail
  
  IBM Spectrum Protect Snapshot Cloning operations can fail if a customer uses 'csh' as login shell for the database user on the cloning target system. Calls to pre- or post-processing scripts during a IBM Spectrum Protect Snapshot Cloning operation will fail with a return code RC=1, and the scripts will not be executed.
  
  Work-around: Switch the database user's login shell on the cloning target system to another shell, for example to ksh or bash. (63869_1)

• Duplicate informational messages
  
  During IBM Spectrum Protect Snapshot operations, a customer may see some duplicate informational messages in the IBM Spectrum Protect Snapshot log files.
  Examples for messages that can occur multiple times:
  FMM6555I Selected snapshot backup with ID '<BACKUP_ID>'.
  FMM6503I Terminating 'fs_chown_clone' request.
  FMM1014I Operation postproc_clone completed successful.
  
  There is no work-around. The duplicate messages can be ignored. (84768, 82060)

• The sequence of a database clone creation without a new database name first, followed by a clone creation with a new database name, fails. The file system mount points / disk group names of the first clone prohibit the mount operation of the second clone that is required before the mount points / disk group names can be changed.
  
  Work-arounds:
• Delete the clone using the production database name before creating a clone with a new database name on the same clone system.
• Use separated clone systems. One for database clones with new database names and one for a database clone using the production database name.

• At least one file system is mounted on the clone target system which is part of a IBM Spectrum Protect Snapshot cloning operation. This means that IBM Spectrum Protect Snapshot will try to mount this file system a second time on the clone system. If running on Linux, this new clone operation will finish successful with the result, that the file system is mounted twice (its strange but its allowed on Unix systems). From an operational perspective the clone works.
  But the delete of the clone using the delete_clone operation does fail. IBM Spectrum Protect Snapshot first unmounts the file system (which was mounted twice). After that IBM Spectrum Protect Snapshot checks if the file system is unmounted but it still detects that it is mounted and fails then with error message
  FMM1241E Unable to unmount the following file systems:
      /oracle/CWS/oradata/HWS
  The file system in question then needs to be unmounted manually or the delete_clone operation needs to be performed again.
  Afterwards a new create clone would work fine and also the delete_clone will work fine.

• The help output of the installer (<installername>.bin --help) includes an option
  
  -r <path_to_generate_response_file> that generates a response file.
  
  The response file generated from the option "-r" cannot be used for silent installation.
  
  Work-around: Use the installer property option ("-DRECORDFILE") to generate an appropriate response file for silent installation. This installer property is explained in the documentation. (73087)

• The command "fcmcli -f delete" only deletes partition 0 of DB2 databases with multiple partitions.
  
  This command, with or without the force command line option '-F', only deletes partition 0 of DB2 databases with multiple partitions. A subsequent request "fcmcli -f inquire[_detail]" does not show the deleted backup any longer. However, the DB2 ACS command db2acsutil query status still shows the backups of the remaining DB2 partitions as valid backups.
  
  General recommendation: To delete DB2 backups with a single partition or with multiple partitions always use the db2acsutil delete command instead of using "fcmcli -f delete".

• IBM Spectrum Protect Snapshot can fail when protecting data stored on DB2 automatic storage in case a data path was specified multiple times when configuring the database.
  A circumvention is documented with APAR IC83834.

• If multiple device classes were used for different partitions for a flashcopy backup in a DB2 DPF environment, IBM Spectrum Protect Snapshot is not able to perform off-loaded tape backups using only one backup server. Instead the following message will be issued:
  FMM6278E No mount agent on a backup server is set up to handle partition '<Nr>' within the current operation.        
  Work-around: Use only one device class for all partitions if possible. If multiple device classes are required, use multiple backup servers which are designated to these device classes.
  
• After installation and configuration of IBM Spectrum Protect Snapshot for DB2 with an off-load configuration using IBM Spectrum Protect for Enterprise Resource Planning the off-loaded backup fails with the following error message:
  db2_all "<<+0<\" echo [ Backup database partition ## ]; db2 BACKUP DATABASE SD1 LOAD /opt/tivoli/tsm/tdp_r3/db2/libtdpdb2.so OPEN 1 SESSIONS OPTIONS @/db2/SD1/tdp_r3/vendor.env exclude logs 2>&1 "
  [ Backup database partition 0 ]
  SQL2071N  An error occurred while accessing the shared library
  "/opt/tivoli/tsm/tdp_r3/db2/libtdpdb2.so". Reason code: "1".
  The shared vendor library /opt/tivoli/tsm/tdp_r3/db2/libtdpdb2.so does not exist with the given name and directory. Instead it is located in /opt/tivoli/tsm/tdp_r3/db264/libtdpdb264.so.
  Work-around: Change the IBM Spectrum Protect Snapshot profile parameter VENDOR_LIB from TSM4ERP to /opt/tivoli/tsm/tdp_r3/db264/libtdpdb264.so or create the following symbolic links:
  cd /opt/tivoli/tsm/tdp_r3
  ln -s db2 db264
  cd db2
  ln -s libtdpdb2.so libtdpdb264.so

• If there are multiple databases managed by a single instance, only one of these databases can be protected by IBM Spectrum Protect Snapshot.
  
• When configuring an IBM Spectrum Snapshot instance the script setup_db2.sh allows to specify the value "GPFS" for parameter COPYSERVICES_HARDWARE_TYPE. However, the value "GPFS" is supported for DB2 pureScale environments only, but not for standard DB2 environments, and specifying this value in a standard DB2 environment results in an inoperable configuration.
  
• Known Issues and Limitation specific to DB2 pureScale
• All limitations specific to IBM Spectrum Scale apply.
• If the backup cluster in a DB2 pureScale environment is distributed across various machines the node where db2 pureScale member 0 is running must be chosen as the backup server for IBM Spectrum Protect Snapshot.
• When configuring IBM Spectrum Protect Snapshot for DB2 pureScale the script setup_db2.sh may fail with the following output:
  
  If the daemons are not started yet:
  Failed to deactivate the IBM Tivoli Storage FlashCopy(R) Manager management daemon. Aborting.
  
  If the daemons are already running (in case of reconfigure):
  /db2/BW1/sqllib//acs/setup_db2.sh: line 370: return: utback3.boeblingen.de.ibm.com:: numeric argument required
  Failed to deactivate the IBM Tivoli Storage FlashCopy(R) Manager management daemon. Aborting.
  
  It is independent if this action is performed as root user or DB2 user.
  
  As a work-around the setuputil script located in the same folder as the setup_db2.sh could be modified as follows:
  Replace line 4924:
      if test $? -eq 0; then
  with
      if test $? -gt 1; then
  
  Result:
  4920 #echo "Deactivating TSM ACS management daemon.";
  4921 echo "";
  4922 print_msg 15052;
  4923 uninstallAgents;
  4924 if test $? -gt 1; then
  4925   #FMM15053T
  4926   #echo "Failed to deactivate the TSM ACS management daemon. Aborting.";
  4927   print_msg 15053;
  4928   DEBUG ENDFUNCTION setup_action;
  4929   exit 2;
  4930 fi
  
  This corrects the main issue. If the daemons are running during execution of the script the following message still appears but the script is able to stop the daemons:
  /db2/BW1/sqllib//acs/setup_db2.sh: line 370: return: utback3.boeblingen.de.ibm.com:: numeric argument required
  Failed to deactivate the IBM Tivoli Storage FlashCopy(R) Manager management daemon. Aborting.
  
  Because the setup script is able to stop the daemons a rerun of the setup_db2.sh call will work fine with this fix.
  This change needs to be deployed in the scripts in the install folder and also in the instance folder.

• User actions necessary in Oracle Real Application Clusters (RAC)
  
  When using IBM Spectrum Protect Snapshot in Oracle Real Application Cluster (RAC) environments make sure that the following actions are performed:
• A database log switch must be performed on all nodes of the Oracle RAC cluster after each IBM Spectrum Protect Snapshot backup.
• Redo logs must be archived on all nodes of the Oracle RAC cluster after each IBM Spectrum Protect Snapshot backup.
• When configuring IBM Spectrum Protect Snapshot in a SAP Oracle environment the action may fail with the following error message:
  FMM8448E Another device class already exists under the given name 'STANDARD'. Please specify another name.
  The problem only occurs if IBM Spectrum Protect Snapshot is configured for cloning only and uses
  the cloning device class with the name STANDARD. When IBM Spectrum Protect Snapshot is reconfigured for backup with off-load and cloning then the configuration wizard tries to add a new device class for backup with the name STANDARD. This results in the above error message and the IBM Spectrum Protect Snapshot profile wizard fails to create the new configuration.
  
  Work-around: Manually duplicate the existing cloning device class in the profile and remove the CLONE_DATABASE YES parameter. If the storage system is SVC and uses the TARGET_NAMING parameter in the FCM profile, then remove the USE_FOR_CLONING <dbname> clause on the TARGET_NAMING parameter as well. Afterwards the wizard starts without any errors.
• When configuring an IBM Spectrum Snapshot instance the script setup_ora.sh allows to specify the value "GPFS" for parameter COPYSERVICES_HARDWARE_TYPE. However, the value "GPFS" is not supported for Oracle (in SAP environments) environments, and specifying this value results in an inoperable configuration.
• Multi-tenant (pluggable) databases are not supported
  

• IBM Spectrum Protect Snapshot database cloning operations are not supported on Oracle Real Application Clusters (RAC) databases - This limitation was lifted in version 4.1.3
• IBM Spectrum Protect Snapshot database cloning operations are not supported on Oracle ASM Database version 11gR1
• User actions necessary in Oracle Real Application Clusters (RAC)
  When using IBM Spectrum Protect Snapshot in Oracle Real Application Cluster (RAC) environments make sure that the following actions are performed:
• A database log switch must be performed on all nodes of the Oracle RAC cluster after each IBM Spectrum Protect Snapshot backup. This limitation was lifted in version 4.1.3
• Redo logs must be archived on all nodes of the Oracle RAC cluster after each IBM Spectrum Protect Snapshot backup.
• IBM Spectrum Protect Snapshot off-load and database cloning operations cannot share one server used as backup server and cloning server for Oracle ASM databases
• Database cloning with Oracle ASM: If the one of the log archive destinations refers an ASM disk group that is not part of the storage snapshot, then the create- or refresh_clone operation will fail. The alert log file will contain a statement like: 
  ARCH: Error 19504 Creating archive log file to '+<disk group name>'
  This limitation also applies if no log archive destination was specified and db_recovery_file_dest refers an ASM disk group that is not part of the storage snapshot. db_recovery_file_dest is used as default log archive destination if not otherwise specified. All disk groups that contain data files, control files or online log files are part of the storage snapshot.
  Work-arounds:
• Set the log archive destination to a disk group that is part of the storage snapshot or set it to a directory within a regular file system.
• In case no log archive destination was specified, set the db_recovery_file_dest to a disk group that is part of the storage snapshot or set it to a directory within a regular file system.
• Create the disk group that is claimed to be missing within the alert log file on the clone system manually
• When installing IBM Spectrum Protect Snapshot 4.1.3 in a language different from English on an Oracle RAC instance, some Installer panels show English messages instead of showing them in the selected language. (103441)
• When installing IBM Spectrum Protect Snapshot 4.1.3 in an Oracle RAC instance, the list of possible candidates for the shared target directory for the installation does not show up.
  The string "NFS and ACFS file systems:" is present, but the list of shared file systems available is not displayed. (103868) This limitation was lifted in version 4.1.4.
  Work-around:
  Either specify the directory to be used for the shared installation manually, or select it via GUI option "Choose dir". The list of eligible shared directories can be obtained via system commands like
  "nfsexport" or "mount".
• When installing IBM Spectrum Protect Snapshot 4.1.3 for Oracle, a generated response file may not work in all environments under some rare conditions. In this case, the silent installation using the flawed response file ends with a "FATAL ERROR" message in file installation.log. (103388)
  Work-around:
  If the described symptom is encountered during silent installation use GUI or console mode installation to install IBM Spectrum Protect Snapshot 4.1.3 for Oracle.
• When configuring an IBM Spectrum Snapshot instance the script setup_ora.sh allows to specify the value "GPFS" for parameter COPYSERVICES_HARDWARE_TYPE. However, the value "GPFS" is not supported for Oracle environments, and specifying this value results in an inoperable configuration.
• Multi-tenant (pluggable) databases are not supported

• On HP-UX with IBM Spectrum Protect Snapshot for Custom Applications and NetApp/N Series storage in NAS mode, a mount operation might fail with the following error:
  
  FMM1232I Executing operation 'mount' on host 'hpback1'.
  FMM6901I Response to Init request.
  FMM1250I The 'root' ID is connecting to the following file server: 'nap02.boeblingen.de.ibm.com' .
  FMM0784I The ONTAP filer version on this appliance is: Data ONTAP Release 7.3.6: Thu Jul 7 02:22:32 PDT 2011 (IBM).
  Not a terminal
  stty: : Not a typewriter
  FMM1220E The execution of the system command '/usr/sbin/mount nap02.boeblingen.de.ibm.com:/vol/RIPhhpprod1dt1_
  TSM_A0HIX44TOR_A0HJ5LTJSP /CAA/RIP/caadata/RIP' ended with a return code other than 0. The output of the command was: 'logout'.
  stty: : Not a typewriter
  .profile[59]: USER: Parameter not set.
  FMM6959I Script 'LD_LIBRARY_PATH=/CAA/RIP/acs/ontapi/lib /CAA/RIP/acs/fmna -l /CAA/RIP/acs -N mount -t -f mount' returned with code 0.
  FMM1233I Operation 'mount' completed.
  
  Work-around: The following lines have to be added in the root user's profile:
  # User defined section:
          #export PS1='\u@\h:\w> '
          export PS1="$LOGNAME@`hostname`:\$PWD>"
  
  (38258)
  
• Starting with IBM Spectrum Protect Snapshot v4.1.4 for Custom Applications, off-loading snapshot backups in IBM Snapshot Scale (a.k.a GPFS) file systems  to an IBM Spectrum Protect server may terminate with message FMM6572E if paths to filesets have a length of about 30 or more characters. In this case, the command output logged with message FMM6572E contains multiple lines
  "awk: String mmbackupIn cannot be longer than 399 bytes.".
  This is a limitation in IBM Spectrum Scale v4.1.1 and v4.2.0 which will be lifted in later deliverables of IBM Spectrum Scale.
  (IBM Spectrum Scale defect 75060)
  
• For Custom Applications on IBM Spectrum Scale or GPFS, all limitations specific to IBM Spectrum Scale apply.

• VIO/vSCSI configuration
  
  For AIX environments that use VIO / vSCSI storage adapter virtualization, the BACKUP_HOST_NAME must be set to the value of PREASSIGNED_VOLUMES when using DS or SVC storage devices.
  Note: The "backup_server_hostname" setting is not allowed.

• When downgrading IBM Spectrum Protect Snapshot for UNIX 4.1.3 to a previous IBM Spectrum Protect Snapshot version on AIX the instance activation (setup_xx.sh -a install -d <instance home directory> -F) aborts during file copy with the following error message:
    cp: not all requested changes were made to /oracle/XS1/acs/jre/jre/lib/ppc64/classic/libjvm.a
    cp: not all requested changes were made to /oracle/XS1/acs/jre/jre/lib/ppc64/j9vm/libjvm.a
    - FAILED: Something went wrong while copying files. Aborting
  The cp command on AIX aborts the file copy because the target of the file copy already exists and is a link.
  Work-around: Check if the files listed in the error output are links pointing to the library in the same folder:
    cd <instance home directory>/acs
    ls -al jre/jre/lib/ppc64/classic/lib
    lrwxrwxrwx 1 oracle oinstall 9 Aug 05 09:18 jre/jre/lib/ppc64/classic/libjvm.a -> libjvm.so
    -rwxr-xr-x 1 root system 284188 Aug 01 01:52 jre/jre/lib/ppc64/classic/libjvm.so
  To be able to run the downgrade of the FCM installation for the selected instance is to delete the jre folder in the acs directory.
    cd <instance home directory>/acs
    rm -rf jre
  Then re-run the instance activation, which will now complete successfully.

• Installation on AIX 7.1 may fail with message 'Error: Port Library failed to initialize' and 'Error: Could not create the Java Virtual Machine.'. See sample below.
  
  ----------------------------------------------------------------------
  # ./4.1.4.0-TIV-TSFCM-AIX.bin -i console
  Preparing to install...
  Extracting the JRE from the installer archive...
  Unpacking the JRE...
  Extracting the installation resources from the installer archive...
  Configuring the installer for this system's environment...
  Launching installer...
  Error: Port Library failed to initialize: -125
  Error: Could not create the Java Virtual Machine.
  Error: A fatal exception has occurred. Program will exit.
  ----------------------------------------------------------------------
  
  Solution: See AIX APAR IV09585. The fix is available in AIX 7.1 TL 1 SP 2. It is recommended to obtain the fix as part of SP2.
  If you cannot upgrade to SP2, you can download the APAR fix available from
  https://www-304.ibm.com/support/docview.wss?uid=isg1IV09585

• Installation when GSKit v8.0.13 is present.
  
  Installing IBM Spectrum Protect Snapshot v4.1 on a Linux machine with GSKit packages of version "8.0.13" already installed will fail with the following error: "Fatal error: No appropriate GSKit installation found."
  rpm -qa | grep gsk will show:
  gskcrypt64-8.0-14.28
  gskssl64-8.0-14.28
  This indicates that the GSKit packages were successfully upgraded to the version delivered with IBM Spectrum Protect Snapshot. However, due to a flaw in the post-install scripts of GSKit 8.0.13, PATH variables are not set correctly so that the GSKit command line tools are not found. The issue was fixed with GSKit version 8.0.13.1.
  
  Work-around: Before installing IBM Spectrum Protect Snapshot, uninstall GSKit version 8.0.13. When IBM Spectrum Protect Snapshot is installed it will also install a more recent GSkit version. (81744)

• SELinux must be disabled
  
  SELinux is activated by default on Red Hat Enterprise Linux. However, SELinux must be disabled for the following environments / operations:
  
  1. SAP/Oracle installations. SELinux should be permanently disabled in this environment. Please also refer to SAP Note 1048303 - „Red Hat Enterprise Linux 5.x: Installation and upgrade“.
  
  2. When installing or uninstalling the IBM Spectrum Protect Snapshot applications, SELinux should be temporarily disabled.
  
  To temporarily disable SELinux before uninstall, enter the following:
  
  setenforce 0
  
  To re-enable SELinux after completing the uninstall process, enter the following:
  
  setenforce 1

• HA features of the supported logical volume managers must not be activated in combination with IBM Spectrum Protect Snapshot V4.1.x.

• Restore May Fail When The LV is Open on Red Hat Enterprise Linux
  
  In rare cases, the Linux LVM does not remove / reset the "Open" state of a logical volume (LV) after the file system for that LV has been unmounted. When this occurs, IBM Spectrum Protect Snapshot restore operations to the affected LV will fail with the following message:
  
  FMM1040E … Remove failed for logical volume <logical_volume_name> ... Can't remove open logical volume <logical_volume_name>
  
  Rebooting the system will reset the state of the LV and you can attempt your IBM Spectrum Protect Snapshot restore operation.

• File systems imported and mounted with IBM Spectrum Protect Snapshot mount operations on Linux are not added to the (v)fstab. Therefore, you would need to remount the file systems manually after a system reboot. This should be considered especially in case of cloned environments that are used over a longer period (and may need to persist beyond machine reboot).
  Work-around: Add the required file systems to your (v)fstab.
  
  
• Linux LVM2 is supported up to lvm2-2.02.115
  Using higher levels will cause operations to fail.
  
  
• Starting with RHEL 7 and SLES 12 LVM2 has enabled the Metadata Daemon (lvmetad) by default. The LVM Metadata Daemon causes Spectrum Protect Snapshot operations to fail.
  Work-around: Disable the LVM Metadata Daemon as described in the Red Hat documentation https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/Logical_Volume_Manager_Administration/metadatadaemon.html

• HA features of the supported logical volume managers must not be activated in combination with IBM Spectrum Protect Snapshot V4.1.x.

• File systems imported and mounted with IBM Spectrum Protect Snapshot mount operations on HP-UX are not added to the (v)fstab. Therefore, you will need to remount the file systems manually after a system reboot. This should be considered especially in case of cloned environments that are used over a longer period (and may need to persist beyond machine reboot).
  
  Work-around: Add the required file systems to your (v)fstab.
• When IBM Spectrum Protect Snapshot on HP-UX is used with IBM System Storage DS8000, IBM System Storage SAN Volume Controller or IBM Storwize V7000, V3700, V5000 and FlashCopy target volumes are configured as preassigned volumes (BACKUP_HOSTNAME PREASSIGNED_VOLUMES) and other target volumes are configured as dynamically assigned volumes (BACKUP_HOSTNAME <storage system hostname>) then FlashCopy mount, off-load or cloning operations on the backup or clone system might fail with the following error message in the summary.log:
  MNT 14:12:46 (35) FMM4003E Operation mount will be aborted. Physical volume(s) with conflicting physical volume id(s) have been found on the target host '<hostname>'
  ae29dedf5249ae91 (/dev/disk/disk101)
  ae29dedf5249ae91 (/dev/disk/disk105)
  
  Work-around: Either use BACKUP_HOSTNAME PREASSIGNED_VOLUMES for all target volumes of all target sets or use BACKUP_HOSTNAME <storage system hostname> for all target volumes of all target sets.
  (44211)
• IBM Spectrum Protect Snapshot database cloning operations of Oracle ASM databases are not supported on this platform
• IBM Spectrum Protect Snapshot off-load and database cloning operations cannot share one server used as backup server and cloning server

• HA features of the supported logical volume managers must not be activated in combination with IBM Spectrum Protect Snapshot V4.1.x.

• Solaris Patch Id 142900-09 must be installed for proper operation of IBM Spectrum Protect Snapshot v4.1 when used with DB2.

• File systems imported and mounted with IBM Spectrum Protect Snapshot mount operations on Solaris are not added to the (v)fstab. Therefore, you will need to remount the file systems manually after a system reboot. This should be considered especially in case of cloned environments that are used over a longer period (and may need to persist beyond machine reboot).
  
  Work-around: Add the required file systems to your (v)fstab.
  
  
• If files or directories are part of the file list IBM Spectrum Protect Snapshot processes within the backup operation that are not residing on file systems resided on disk groups managed by Veritas Volume Manager the following error will occur:
  FMM1040E An exception occurred in the operating system adapter component, exception text: Type: '<class 'xsf.host.storage.filesystem.errors.FileSystemInstantiationError'>', Value: '', Traceback: '<traceback object at 0x23062b0>'
  
  Work-around: Only include files and directories in the backup that are resided on disk groups managed by Veritas Volume Manager.
• The same auxiliary host could be configured as a clone server and a backup server for mounting and unmounting backup images. For example, when you want to off-load snapshots to IBM Spectrum Protect . When a clone image exists and a backup image is mounted on the same server, the IBM Spectrum Protect Snapshot unmount process can fail with the following error message:
  FMM6534E Another snapshot backup is already mounted. 
  
  Work-around: Do not use the same auxiliary host as a backup server and a clone server.
  (41734)

• IBM Spectrum Protect Snapshot database cloning operations of Oracle ASM databases are not supported on this platform

• FCM operations in VMware® Virtual Machines
  IBM Spectrum Protect Snapshot can back up, restore, and perform clone operations on VMware virtual machines that run Linux guest operating systems.
  
  All source LUNs that participate in backup or clone operations must be attached to the virtual machine by using one of the following attachment methods:
  • VMware physical mode raw device mapping (pRDM)
  • iSCSI
  • Network File System (NFS)
  
  IBM Spectrum Protect Snapshot can perform a FlashCopy restore operation from a snapshot to an existing pRDM disk but does not create a virtual machine or pRDM definition as part of the restore process.
  
  If the backup or clone system is a virtual machine then all target LUNs that participate in backup or clone operations must be attached to the virtual machine by using one of the following attachment methods:
  • iSCSI
  • Network file system (NFS)
  
• FCM operations in Linux KVM Virtual Machines
  IBM Spectrum Protect Snapshot can back up, restore, and perform clone operations on Linux KVM virtual machines that run Linux guest operating systems.
  
  All source LUNs that participate in backup or clone operations must be attached to the virtual machine by using one of the following attachment methods:
  • Block device mapping (BDM)
  • iSCSI
  • Network File System (NFS)
  • PCI Passthrough
  
  For the Tivoli® Storage FlashCopy® Manager KVM setup using Block Device Mapping (BDM), the LUNs must be mapped to the KVM guest as multipath devices. The LUNs must be visible inside the KVM guest as multipath devices.
  
  From within the KVM guest, run the multipath command. The output from running this command looks similar to the following example:
  
  kvm-guest:~ # multipath -ll
  mpathat (360050768018205de4000000000001949) dm-7 IBM    ,2145
  size=2.0G features='1 queue_if_no_path' hwhandler='0' wp=rw
  `-+- policy='service-time 0' prio=50 status=active
  `- 3:0:0:3 sdf 8:80 active ready running
  
  The LUN identifier is a unique number like this 360050768018205de4000000000001949, and must not be overwritten by the KVM stack. The product storage identifier like this IBM ,2145 must be visible inside the KVM guest.
  
  If the backup or clone system is a virtual machine then all target LUNs that participate in backup or clone operations must be attached to the virtual machine by using one of the following attachment methods:
  • iSCSI
  • Network file system (NFS)

• Clean-up of the takeover host after remote restore on IBM XIV® Storage System
  
  When using IBM Spectrum Protect Snapshot v4.1 in combination with the remote mirror functionality of an IBM XIV® Storage System or with the metro mirror or global mirror functionality of the IBM System Storage SAN Volume Controller or IBM Storwize V7000, take care to clean up the take-over host after a remote restore before making the original production site the master of the remote relation again.
  The following scenario may leave the instance on the take-over host in an unpredictable state:
1. A remote FlashCopy or snapshot backup is restored to and mounted on the take-over host, for example due to disaster or maintenance.
2. In order to switch back to the original production system, the remote copy relationship is reverted to the original production site as the master.

Work-around: Prior to switching back to the original production site, deactivate the remote backup on the take-over host by performing the following procedure:
1. Unmount all file systems of the backup restored.
2. Vary off-line the volume groups involved.
3. Export the volume groups involved.
4. Remove the devices involved from the take-over host.
(84992)
• Limitation concerning remote relations on IBM XIV® Storage System after restore
  
  When using IBM Spectrum Protect Snapshot v4.1 in combination with the remote mirror functionality of an IBM XIV® Storage System, XIV demands that IBM Spectrum Protect Snapshot deactivates and deletes the remote relationship before restoring a snapshot to either the master or the slave volume of a remote mirror relation. After restoring, IBM Spectrum Protect Snapshot does not recreate the mirror relations.
  
  Work-around: After IBM Spectrum Protect Snapshot restore, use the XIV XCLI or GUI to create the remote relationships and start the synchronization. (87292)
• Backups might fail if a volume migration of OLVM volumes is ongoing. IBM Spectrum Protect Snapshot is able to protect OLVM volumes before and after their migration. Before the migration it is configured to use the source XIV system as storage device. Once the disks were migrated, IBM Spectrum Protect Snapshot needs to be reconfigured to use the target XIV system as storage device for the backup.
  
  
• Limitations related to XIV Multi-Site Mirroring environments:
• FlashCopy Manager fails to restore snapshots if the volumes containing the application data are mirrored into two or more destinations with XIV Multi-Site Mirroring.
• FlashCopy Manager fails to create snapshots of mirrored volumes if the application data is mirrored into two or more destinations with XIV Multi-Site Mirroring. Snapshots of local source volumes can be created.
In both scenarios FlashCopy Manager fails with an error message containing this text: "VOLUME_HAS_MULTIPLE_MIRRORS volume has multiple mirrors, operation not allowed or target must be specified"

Work-around: Remove all mirror relations manually, before triggering a restore operation with FCM.
To create a snapshot of a remote mirror, ensure that only one mirror of the volumes containing the application data exists.

• Limitation for environments where source volumes of a FlashCopy are mirrored by using the ability of the SVC to create volumes with two copies, and where the copies are in two different SVC storage pools: When using dynamic allocation of FlashCopy target volumes in such environments the storage pool for the target volumes of the FlashCopy is not automatically determined.
  Work-around: Specify the SVC storage pool for the target volumes explicitly via parameter SVC_POOLNAME in the DEVICE_CLASS section of the IBM Spectrum Protect Snapshot profile.
  
  
• For IBM Spectrum Protect™ Snapshot to work in combination with dynamic target allocation (SVC DTA) device classes, the name of the SAN Volume Controller production source volumes must not exceed the length of 27 characters.
  
  An IBM Spectrum Protect Snapshot backup involving a volume with a longer name fails with the following SVC error message:
  CMMVC5738E The argument contains too many characters.
  
  There is no work-around to allow source volume names of lengths larger than 27 characters.
  The example below shows however that a volume name smaller than 27 characters can be practicable.
   
  Example:
  Given the sample input source volume name „esxcet1_vol1“, IBM Spectrum Protect Snapshot SVC DTA adapter generates the following object names in SAN Volume Controller:
  
  Name of the dynamically allocated target volume for the snapshot backup operation:
  esxcet1_vol1_A0ITWN5DGH_ITWN6A1F_1
  Name of the duplicate target volume used for the mount operation:
  esxcet1_vol1_A0ITWN5DGH_ITWN6A1F_1_dup
  Name of the FlashCopy mapping:
  TSM_A0IO1BN9X6_A0IW39B9GG_IW39J8A6_esxcet1_vol1
  Name of the consistency group:
  TSM_A0IO1BN9X6_A0IW39B9GG_IW39J8A6

• Misleading error message when requesting additional incremental backups on DS8000
  
  On an IBM System Storage DS8000 series, only a single incremental relation can be created. When a second incremental relation is requested for a DS8000, the following error message is issued:
  
  FMM0777E An unexpected error was encountered processing a function.
  function name : DsHWInterface::hwStartCopy
  function : CIM Error. Failed to execute IBMTSDS_StorageConfigurationService::AttachReplicas , Return: Failed
  CIM Error Code : ReturnCodes: <0xb040 0xb040>
  0xb040: FlashCopy operation failure: action prohibited by current FlashCopy state. Contact IBM
  technical support for assistance.
  return code : 4
  
  Though the error may be misleading, the issue at hand is that it is not supported to create multiple incremental FlashCopy relations on the DS8000 from the same source volume.

• IBM Spectrum Scale snapshots are located in the productive file systems. Due to their copy-on-write technology, their sizes grow over time as the database is modified. When configuring IBM Spectrum Protect Snapshot, administrators are asked to assess how many file system snapshots can be kept in the file system without risking a file-system-full condition.
  The number of IBM Spectrum Scale snapshots kept in the file systems is controlled with IBM Spectrum Protect Snapshot profile parameter MAX_VERSIONS. This parameter cannot be set to the value ADAPTIVE in IBM Spectrum Scale environments, but must be set to the number of snapshots to be kept.
  During production, administrators must continuously monitor free space in the IBM Spectrum Scale file systems where the database and log files reside in order to avoid a file-system-full condition of the productive file systems. Monitoring can be done with commands df and mmdf.
• Nested IBM Spectrum Scale filesets are not supported.
  Specifically, if the root fileset of a file system participates in the backup other filesets are not allowed in this file system. If the root fileset does not participate in the backup various independent filesets linked in the root fileset can be defined.
• Nested file systems (i.e. file systems mounted on paths in other file systems) are not supported.
• Symbolic links are supported only when the links point to files in the same file system and fileset.
  
  

• Backup of OLVM targets is not possible.
  Preconditions of this problem:
  - The disks to protect have a configured OLVM (OnLine Volume Migration) target
  - The disks were migrated to the OLVM target
  
  Before the volume migration the backup works fine. IBM Spectrum Protect Snapshot is configured to use the source XIV system as storage device. Once the disks were migrated and FlashCopy Manager was reconfigured to use the target XIV system as storage device the backup fails with this message:
  FMM6981E Received an error from the adapter: The LUN with serial number <serial> is not known by the configured CIM Agent <IP> of storage system type "XIV"

• IBM Spectrum Protect Snapshot database cloning operations are not supported on Oracle Real Application Clusters (RAC) databases
  
• User actions necessary in Oracle Real Application Clusters (RAC)
  When using IBM Spectrum Protect Snapshot in Oracle Real Application Cluster (RAC) environments make sure that the following actions are performed:
• A database log switch must be performed on all nodes of the Oracle RAC cluster after each IBM Spectrum Protect Snapshot backup.

• When installing IBM Spectrum Protect Snapshot 4.1.3 in an Oracle RAC instance, the list of possible candidates for the shared target directory for the installation does not show up.
  The string "NFS and ACFS file systems:" is present, but the list of shared file systems available is not displayed. (103868)
  Work-around:
  Either specify the directory to be used for the shared installation manually, or select it via GUI option "Choose dir". The list of eligible shared directories can be obtained via system commands like
  "nfsexport" or "mount".