Maintenance Pack NB_45_9_M.sgi.tar provides fixes for VERITAS NetBackup (tm) SGI IRIX servers. NetBackup UNIX Add-on products, Database Agents and Java GUI have separate Maintenance Packs.

Article:TECH43741  |  Created: 2005-01-20  |  Updated: 2013-10-23  |  Article URL http://www.symantec.com/docs/TECH43741
Article Type
Technical Solution

Product(s)

Environment

Issue



Maintenance Pack NB_45_9_M.sgi.tar provides fixes for VERITAS NetBackup (tm) SGI IRIX servers. NetBackup UNIX Add-on products, Database Agents and Java GUI have separate Maintenance Packs.

Solution



NB 4.5MP6 Pack NB_45_9_M README September 20, 2005
Corequirement: NB_CLT_45_9_M
================================================================================
This Maintenance Pack provides fixes for VERITAS NetBackup (tm) UNIX servers.
NetBackup UNIX Add-on products, Database Agents and Java GUI have separate
Maintenance Packs.

Symantec recommends that the backing up of active file systems be avoided,
or the use of snapshot technologies be implemented. The directory structure
reported back from the file system (to NetBackup) may not contain all of the
files available during the time of backup. NetBackup will not report errors
in many cases where the file's existence is not known to NetBackup as reported
by the file system.

================================================================================



=================
PACK DEPENDENCIES
=================

-- 4.5 MP6 must be installed prior to installing this Maintenance
Pack.

-- Installation of this Maintenance Pack requires version 1.57 of the
Vrts_pack.install script.

-- For the latest robotics support please also download and install the
latest Mappings_4.5.<6 digit number>.tar from the support Web site:
http://www.support.veritas.com .


I. DOWNLOAD INSTRUCTIONS
II. KNOWN ISSUES
III. INSTALLATION INSTRUCTIONS
IV. UNINSTALL INSTRUCTIONS
V. DESCRIPTION OF PROBLEMS FIXED
Current Pack
NB_45_7_M
NB_45_8_M

=========================
I. DOWNLOAD INSTRUCTIONS
=========================
1) Download the NB_CLT_45_9_M_<6 digit number>.tar and
NB_45_9_M_<6 digit number>.<server>.tar files into the
/tmp directory,

where <6 digit number> is an internal tracking identifier

where <server> is alpha, hp_ux, linux, ncr, rs6000, sequent4.2, sgi, solaris

NOTE: NB_CLT_45_9_M_<6 digit number>.tar has the client binaries and
NB_45_9_M_<6 digit number>.<server>.tar has the server binaries and BOTH
must be installed.

2) Extract the NB_CLT_45_9_M_<6 digit number>.tar and the
NB_45_9_M_<6 digit number>.<server>.tar files.
tar xvf NB_CLT_45_9_M_<6 digit number>.tar
tar xvf NB_45_9_M_<6 digit number>.<server>.tar

NB_45_9_M will create the files:
VrtsNB_45_9_M.README
VrtsNB_45_9_M.<server>.tar.Z
VrtsNB_45_9_M.postuninstall
VrtsNB_45_9_M.postinstall
VrtsNB_45_9_M.preinstall
Vrts_pack.install

NB_CLT_45_9_M will create the files:
VrtsNB_CLT_45_9_M.README
VrtsNB_CLT_45_9_M.tar.Z
VrtsNB_CLT_45_9_M.postuninstall
VrtsNB_CLT_45_9_M.postinstall
VrtsNB_CLT_45_9_M.preinstall


=================
II. KNOWN ISSUES
=================
The following are known issues with NB_45_9_M. Please note that
these known issues will NOT be resolved after installing this pack.

Description:
The vxlicense utility and elm licenses that are used for VxFS 3.3 and
VxFS 3.4 are no longer used with VxFS 3.5. VxFS 3.5 uses a new license
utility and new licenses. Customers using NBU BLIB on Solaris or HP-UX
systems will experience a problem causing BLIB to fail if the system
is a clean install of VxFS 3.5. There is no problem if this was an
upgrade from VxFS 3.3 or 3.4 up to VxFS 3.5.

Workaround:
Install the old vxlicense utility on the system and install the old
elm licenses using that utility.

Description:
The "backuptrace" command of the log analysis utility, introduced in
this maintenance pack, will hang on Tru64 systems. Avoid using this
command on Tru64 systems.

Description:
If NetBackup is installed on a VxFS file system (VxFS 3.4) on AIX 5.1,
the system may stop responding when the NetBackup daemons are started.
This problem is isolated to AIX 5.1 and VxFS 3.4 with NetBackup installed
on a VxFS file system.

Workaround:
The minimum AIX kernel patch level for supporting VxFS on AIX 5.1 is:

1. AIX 5.1 at ML02 with APAR IY32907, or
2. AIX 5.2 at ML03

The APAR IY32907 is required at ML02 level to prevent AIX kernel from
crashing while loading/executing certain NBU daemons that reside in
vxfs file systems. This crash only occurs when the target file system(s)
for installing NBU binaries was vxfs. If target file system for NBU was
not vxfs, the crash does not occur. Regardless of the type of file system
used for installing NBU, this APAR is required for correct operation of
VxFS on AIX 5.1 at ML02.

Description:
After upgrading to NetBackup 4.5 Maintenance Pack 2, AIX client restores
are now getting exit status 5.

If AIX is using VxFS, the Master Servers, Media Servers and Client
Backups/Archives/Restores will work fine if all platforms are at the
Maintenance Pack 2 level. If the Client is at Maintenance Pack 2, but the
Master and/or Media Servers are still at GA level, backups will work fine,
but restores will fail with an error code 5.

Workaround:
Upgrade the Media and Master Servers to Maintenance Pack 2 and retry the
restore.

Description:
The installation of the NetBackup client Maintenance Pack, NB_CLT_45_2, may
fail on HP-UX machines under the following circumstances:
Sybase IQ 12.4.3 (32 bit) is running, and a backup using the NetBackup for
Sybase agent has been previously done.

When doing a local installation of the NetBackup client Maintenance Pack,
the installation may fail with the following error:

mv: /usr/openv/lib//libVcvcomb.sl: cannot write: Text file busy
mv: /usr/openv/lib//libxbsa.sl: cannot write: Text file busy

When doing a remote installation of the NetBackup client Maintenance Pack,
the installation may fail with the following error:

Couldn't open /usr/openv/lib/libVcvcomb.sl on client
Client open errno = 26

Workaround:
The workaround for this problem is to stop Sybase IQ, and then to do the
installation of the NetBackup client Maintenance Pack again.

**Description:
NetBackup for Lotus Notes does not support the file list directive,
ALL_LOTUS_DATABASES.

Workaround:
Add the Domino data directory, containing the database files with
extensions .nsf, .ntf or .box, in the file list instead of using the
ALL_LOTUS_DATABASES file list directive. There will be an increase in
performance since the NetBackup for Lotus Notes extension does not have
to parse the file system for .nsf, .ntf or .box extensions.

Description:
NetBackup extension for Lotus Notes is not supported in the
Microsoft Cluster environment (MSCS). NetBackup does support Lotus Notes
in a non-clustered configuration.


===============================
III. INSTALLATION INSTRUCTIONS
===============================
As root on the NetBackup Master/Media Server:

1) Install NB_45_9_M and NB_CLT_45_9_M Maintenance Pack binaries.

cd /tmp
/bin/sh Vrts_pack.install

2) Restart daemons.

/usr/openv/netbackup/bin/initbprd
/usr/openv/volmgr/bin/ltid -v

3) Update the NetBackup clients, including the NetBackup master and media
servers, with the update_clients script.

/usr/openv/netbackup/bin/update_clients <hardware> <os>

where <hardware> <os> is one of the following:
ALPHA OSF1_V4
ALPHA OSF1_V5
DataGeneral UNIX
HP9000-700 HP-UX11.00
HP9000-800 HP-UX11.00
INTEL FreeBSD
Linux RedHat2.2
Linux RedHat2.4
MACINTOSH MacOSXS1.2
MACINTOSH MacOSX
NCR UNIX
RS6000 AIX4.3.3
RS6000 AIX5
SCO UnixWare
Sequent DYNIX420
SGI IRIX65
Solaris Solaris2.6
Solaris Solaris7
Solaris Solaris8
Solaris Solaris9
Solaris Solaris_x86_2.6
Solaris Solaris_x86_7
Solaris Solaris_x86_8

Remember to include the master server's <hardware> <os> type.

Note: The /usr/openv/netbackup/bin/update_clients command without
any parameters will update all the UNIX clients.

Note: When updating an RS6000 client, there may be circumstances
where update_clients will fail with an error similar to
this:

Couldn't open /usr/openv/lib/libVmangle.so on client
Client open errno = 26

If this happens, execute /usr/sbin/slibclean on the client
to be updated and re-run update_clients.

If the client (CLT) .Z file and README exist in the installation
directory during the installation of the server maintenance pack,
the Vrts_pack.install script will install the client maintenance
pack automatically. The client maintenance pack will NOT be
installed automatically during a reinstall of the server
maintenance pack.

Additional Notes:

If non-root administrators use the GUI only, the nonroot_admin
script no longer needs to be run. If the non-root administrators
use the command line or bpadm, the group and file permissions
will have to be changed manually on the NetBackup binaries.
Users can writer their own script. The script is being phased
out because there is a slight security risk that non-root users
may be able to execute NetBackup commands only because those users
are part of a group that is allowed to execute NetBackup commands.


===========================
IV. UNINSTALL INSTRUCTIONS
===========================
Note: This will ONLY uninstall the Maintenance Pack from your local machine.

1) Close the NetBackup user interfaces.

Make sure the NetBackup server has no active jobs running (for
example, backups, restores, or duplications).

If a database agent is being used, such as Oracle,
ensure that the database services are stopped.

2) Change directory to the patch save directory.
Substitute the pack name for ${PACK} in the following command:

cd /usr/openv/pack/${PACK}/save

3) Run the un-install script:

./Vrts_pack.uninstall

4) Verify that the pack uninstalled successfully by checking
/usr/openv/pack/pack.history.

5) If update_clients was run after the pack was originally INSTALLED,
run it again after that pack is successfully UNINSTALLED.

6) If necessary, restart the NetBackup and Media Manager daemons:
/usr/openv/netbackup/bin/goodies/netbackup start


=================================
V. DESCRIPTION OF PROBLEMS FIXED
=================================
The following are descriptions of the problems fixed.
Please read the entire document before installing.

README Conventions:

Description
Describes particular problem or feature contained in this Maintenance Pack.

** Description **
Describes a problem that can lead to potential data loss. Please
read these problem descriptions carefully.

Workaround
Any available workarounds to a problem are also listed. Workarounds
can be used INSTEAD of applying the patch, however, Symantec strongly
recommends the "best practice" of being at the latest patch level.

Additional Notes
Any additional information regarding this problem or feature is included.


=============
Current pack
=============

================================================================================
Description:
Message lines in the LTIerrno.h file were not found by the search program
if the message lines began with spaces instead of a tab.

Additional Notes:
The list of messages for Media Manager errors has changed. This only
effects localized sights.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The scheduler was modified to correctly process a status 134 error, race
condition in a multiplex backup.

This condition occurs when a multiplex child, bpsched, starts a backup, but
the storage unit is over committed. A status 134 error is sent back to the
multiplex child, bpsched, but the child sched is already adding another
backup to the multiplex group.

The child sched attempts to send the new jobs information to bpbrm, but
bpbrm has exited and the connection is defunct. The scheduler was
incorrectly logging the defunct connection in the job progress log. With
this change scheduler no longer logs the broken socket, and correctly reads
and processes the status 134 error from the inbound connection.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
bpretlevel loops echoing the word, "What?" when changing retention levels.
Entering a ctrl-C command is the only way to exit this loop.

An earlier code change had changed all occurrences of gets() to fgets().
However, this change failed to recognize that fgets() return the "\n",
where the previous code, using "gets()" did not. The subsequent strcmp
was not expecting the "\n", and caused the failure. To resolve this issue,
the instances of the [new] trailing "\n" were stripped to emulate the
"gets()" behavior.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Multiplexed backup jobs would occasionally hang. This would happen if the
bptm or bpdm parent process crashed, or if bptm could not successfully add
the TIR fragment for a TIR backup. It might also happen because of a
timing problem in bpbrm when new jobs were started.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Corrected an issue that caused the scheduler to lose a backup in the
middle of a catalog backup.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Backups of NFS mounts would timeout if the NFS server is down or in
single-user mode.

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
The VxFS team changed their implementation of ACLs on VxFS file systems on
the AIX platform. NetBackup was changed to handle this new implementation.

(NetBackup Clients: RS6000_433 RS6000_51 )
--------------------------------------------------------------------------------
Description:
Whenever the vmd or any of its children terminated, it would log the
error code, "terminating - <error_string> (< error_code> )" to the syslog.

Many times, the error code that was logged was, "terminating - successful
(0)", which is a normal exit status. This change eliminates logging a
normal exit status such as this to the syslog.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
bpdbm would crash if it was unable to determine the host name of caller.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When bpbrm encountered an error during backup, bpbkar did not shut down
properly. On HP11.11 platforms, bpbrm would infinitely loop until an
alarm signal would go off after 300 seconds and cause millions of debug
messages to be written to the debug log. On other UNIX platforms, bpbkar
would shut down abnormally because of a segmentation fault.

Workaround:
To avoid this issue, turn off bpbkar debug logging on HP11.11 systems.

(All NetBackup UNIX Clients )
--------------------------------------------------------------------------------
Description:
When bpbackup is run with the option to wait for the completion of the
backup job, a firewall between bpbackup and bprd can shut down the socket
while bpbackup waits for the completion status. This causes backint to
get stuck in an infinite loop during database backups.

(All NetBackup Windows Clients All NetBackup UNIX Clients )
--------------------------------------------------------------------------------
Description:
An XBSA application would hang during a shared memory backup (such as, the
client and the media server are the same machine) if a problem occurred
that caused the bptm to exit. The BSASendData() call did not return.

(All NetBackup Windows Clients All NetBackup UNIX Clients )
--------------------------------------------------------------------------------
Description:
The catalog cleanup process would leave ASCII .f files for Flashbackup
images open. This led to the failure, "too many open files".

Workaround:
Manually running clean-up one client at a time can help resolve this
problem. This will also split file lists into multiple policies.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
If the bpgetmedia options -h and -M were both specified, the -h argument
would always inherit the -M argument, and the original -h value would be
lost.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
NDMP restores fail with Hitachi NAS when restoring the entire backup image.
The NDMP log shows the path being sent twice in the NDMP_DATA_START_RECOVER
message.


Workaround:
Restores will work as long as the file system path is not selected (the
path that was initially backed up). If the entire image is needed for
restore, select all the files and directories in the file system path
instead.

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
NetBackup processes may core dump if an error occurs. The stack from the
core file showed problems with the ndmp_user_scsi() or ndmp_user_tape()
functions.

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
Added a new ID for two duplicated messages.

(All NetBackup Servers)
================================================================================

=========
NB_45_8_M
=========

Description:
The scheduler's multiples group list on Windows 2000 platforms was being
truncated.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Bprd hits a segmentation fault and core dumps. The issue occurred when the
log-file path for the bprestore log exceeded 58 characters. This only
occurred when the media server involved was not the master server, and when
FQDN names were used.

To resolve this issue, the command buffer size has been increased to
BUFFLEN (4096), and implemented bounds checking on all string buffers,
where practicable, by using V_strlcpy, V_strlcat, and snprintf().

Additional Notes:
While libVcom.so and netbackup.dll contain no changes, they are required
to resolve new entry points introduced in libsfr.so. In addition,
libsfr.so is Solaris only.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When scheduler verbose was set greater than zero, Jobs could exit with a
status 50. This problem was more likely to occur on systems experiencing
connection issues. The problem occurred when the scheduler printed a
message and lost the errno value. Armed with the incorrect errno value,
the scheduler would exit (status 50) the current running job(s).

Workaround:
To avoid the issue, set scheduler verbose to zero.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When "Browsing for restore", files that ended in an asterisk (*) or an
angstrom (@) would have that particular symbol (* or @) truncated. This
prevented them from being restored.

(All NetBackup UNIX Clients )
--------------------------------------------------------------------------------
Description:
XBP was inserting and deleting items from lists one item at a time. This
was done because of deficiencies in some X-Servers' support of 1.1 Motif.
This is not an efficient way to handle lists with a large number of items.
This problem is not an issue with X-Servers that support Motif 1.2.

(All NetBackup UNIX Clients )
--------------------------------------------------------------------------------
Description:
The bpplschedrec -deldayomonth command deletes a range of days instead of
the specified day.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Corrected potential problems with memory leaks and non-initialized malloc
space.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
A backup is established with:
- multiplexing enabled
- multi-copy enabled
- continue backup on copy failure

Excessive failure messages were displayed if one of the copies fails, and
additional backups attempt to join the same multiplex group.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The interrupt code paths for the Scheduler were simplified to eliminate
scheduler core dumps because the interrupt handlers were too "noisy".

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Enhance bptm to use MEDIA_UNMOUNT_DELAY for duplications to reduce
unnecessary unmounts/re-mounts during Vault sessions.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
"Ghost" backup jobs were appearing in the Activity Monitor. A deleted
backup job could (partially) re-appear. This would only happen to a deleted
job that was one number lower than a multi-copy backup.

For example:
1) Job n finishes.
2) Job n is deleted from the activity monitor.
3) Multi-copy job n+1, n+2, n+3 finishes.
4) Activity monitor incorrectly updates jobs n, n+1, n+2.
5) Deleted job n (partially) re-appears.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Oracle recovery templates caused a core-dump if it was used/managed under
Template Administration window.

In addition, the Oracle Backup wizard enabled a recovery template to be
selected when choosing the option of "Retrieve configuration of an existing
template".

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The scheduler does not automatically start "bpexpdate -deassignempty" to
deassign media that do not have associated images after "bprecover -l" has
been run.

Workaround:
If a bprecover -l has been executed, remove the following file:

UNIX:
/usr/openv/netbackup/bin/deassign_lock

Windows:
VERITAS\NetBackup\Bin\deassign_lock

Do not remove this file if a catalog recover (bprecover -r) has been done.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
A loss of multiplexed groups within bpsched.exe resulted in the failure to
cancel multiplexed backups.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Single file restores from the flashbackup of multiple raw-partitions can
fail.

(All NetBackup Servers: Solaris)
--------------------------------------------------------------------------------
Description:
You can now view the correct number of files backed up in the Activity
Monitor.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
During an alternate client restore of Informix, the physical restore is
successful. However, the logical log restore always fails while restoring
the last log requested. This happens whether the restore is a full
restore, point-in-time restore, or a point-in-log restore.

The transaction appears to be complete on the client and the client
disconnects the socket connections to the media server. The media server
is not expecting this and exits with status 24.

(All UNIX NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Backups with exclude and include lists could incorrectly back up excluded
directories (but not the files within those directories).

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
During multiplexed restores, when multiple tar instances are restoring
files onto the same directory tree, some of the files may not get restored.

(All NetBackup UNIX Clients )
--------------------------------------------------------------------------------
Description:
When dealing with file sizes less than a block, a compression operation
would return a status 256. The bpdbm log would then show the following
when trying to browse for files in the restore browser:

17:05:07.264 [40258] < 16> decompress_dbfile: system()
returned 256 (1) 17:05:07.577 [40258] < 16> list_files: cannot
decompress /usr/openv/netbackup/db/
images/alex/1072000000/Production_Windows_1072660144_FULL.f.Z

Workaround:
To avoid this issue, ignore the error message in the log.

(All NetBackup Servers: Alpha)
--------------------------------------------------------------------------------
Description:
SCSI-3 library commands have standardized on op code 0x37 for the command
INITIALIZE ELEMENT STATUS w/RANGE. Previous implementations by numerous
vendors had used vendor-unique code 0xE7 - which was implemented by
NetBackup. SCSI-3 libraries will return sense data 20/00 when sent an E7.

Workaround:
To avoid this issue, use only tapes with barcodes that are valid for your
library.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
NetBackup was not properly examining the SCSI library mailslot status bits;
therefore, during ejects of more than a MAP-full of media, the eject
operation may abort after the first MAP-full is removed.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Some TL8 robots that would eject more than a MAP-full (media access port)
of media would abort after the first MAP-full.

Workaround:
To avoid this issue, re-issue the eject operation with the remaining media
to be removed from the robot.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
To assist image database cleanup in 24x7 environments, image database
cleanup will always run in background. For example, bpsched will not
wait until the cleanup is complete before scheduling new jobs. In
addition, bpsched will initiate an image database cleanup at periodic
intervals, 12 hours by default. The time interval may be changed by using
the bpconfig -cleanup_int <hours> command. For example, to change the
interval to 24 hours, execute the following command:

bpconfig -cleanup_int 24

Use a value of 0 to disable bpsched from starting the periodic image
database cleanup. The maximum value is 744 hours. Bpsched will still
perform image database cleanup at the end of a session, however, it will be
run in background. On UNIX systems, bpadm may also be used to change the
interval.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
A large number of down storage units would result in and apparent
scheduler hang. In reality, the scheduler was timing out to each media
server in series. This fix recognizes and skips duplicate attempts to a
down media server.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The predict algorithm was not sorting correctly. It failed to position
cumulative incremental backups before the differential incremental backups.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When selecting a media ID for backup, if the allocation date in the
media database does not match the assigned date in the media manager
database, FREEZE the media because there could be database inconsisties.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The bpsched log would fill with the following message:

"storage unit was not assigned for clientjob ..."

This was caused by an erroneous invocation of the add to multiplexed group.
The add to multiplexed group function would performed a parameter
validation, print the message, and return.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When running VSP-enabled backups of the Veritas NetBackup install drive
(typically C:\) on the master server, VSP snapshot creation consistently
fails for all backups that are run on the VERITAS NetBackup installation
drive. Backups like this would end with status 1 because the VSP snapshots
could not be enabled to allow the backup to back up open/active files.
This means the open/active files were inaccessible to the backup. This
issue only affects the VSP snapshot creation of the NetBackup install
directory and not any other drive on the master server host.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Bpbrm received a SIGSEGV and core dumped in the malloc system call. This
issue has been resolved.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The scheduler would take a long time scheduling multiple data stream jobs
if there were many policies that were due to be autodiscovered at once.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
In a rare circumstance, the last block of a multiplexed, duplicated copy of
a TIR backup may be corrupt. A verification of the copy would not be
successful. The original backup is not affected. This does not affect
non-multiplexed duplications of a multiplexed TIR backup.

(All NetBackup Servers)
================================================================================



=========
NB_45_7_M
=========
Description:
When using the ftp or rsh methods to push MACINTOSH Mac OS X 10.2
client software to Macintosh machines running 10.3, the install
would fail. This problem has been fixed.

(NetBackup Clients: MacOSX macintosh Macosxs )
--------------------------------------------------------------------------------
Description:
Offhost backup on HP-UX systems, including Media Server Copy and Third
Party
Copy, fails.

The bpbkar log shows the following error:

10:28:09.509 [4098] <16> bpmap_send_extend: ERR - tpc_get_disk_info()
failed: err 11
10:28:10.085 [4098] <16> bpbkar Exit: ERR - bpbkar FATAL exit status = 11:
system call failed

(NetBackup Clients: HP SOLARIS )
--------------------------------------------------------------------------------
Description:
When starting several mounts simultaneously, the user may see "unable to
read ack" messages in the syslog from the robot daemon as it tries to
communicate to the robot control daemon. This is more likely to occur with
TL8 libraries, and can occur with either. This may also lead to a hung
parent daemon that reports a syslog error of being out of file handles.
(errno = 24, EMFILE)

TLD only: NetBackup re-initializes the ADIC Scalar 10K library before
downing the drive in the event of a hardware error on the drive. NetBackup
will no longer reinitialize this library.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When doing a detailed import or verify, when there are lots of small
files, the logging takes longer than it should.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
In some restores from multiple images on the same tape, more than one
image restore can be started at the same time causing bptm to core dump.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
If a non-VxFS file system were mounted under a VxFS file system, ACLs would
not be read correctly for edi directory entries. This change brings the
code into alignment with the FP code. This problem only occurs on AIX
platforms.

(NetBackup Clients: RS6000_433 RS6000_51 )
--------------------------------------------------------------------------------
Description:
When restoring large numbers of small files with a detailed progress log
on a media or master server, the O_SYNC flag on the open of the progress
log can cause the restore to take much longer than it should.

Workaround:
Do the restore from the client or don't get a detailed progress log.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The bpbrm -stream_count argument is not correct for frozen image or split
mirror backups. The error manifests itself in the "-stream_count" being
zero for stream number 1, and 'n-1' for streams 2 through n.

It's worse for frozen image or split mirror backups using retention level
24: the -stream_count argument is zero for all streams.


Workaround:
1) Stagger multi-stream frozen image job start windows.
2) Avoid retention level 24 for multi-stream frozen image jobs.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Extended copy backups may fail on HP UX media servers (using an HP Tachyon
fiber channel HBA) due to a tape position miscompare. The tape is then
frozen and the backup fails with an error code 174.

(All NetBackup Servers: HP)
--------------------------------------------------------------------------------
Description:
If a file has VxFS attributes such as reserve or fixed extent size, a
FlashBackup file restore would fail resulting in a zero length file.

(All NetBackup Servers: HP RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
When a backup is terminated using the job monitor, bpbkar may read the
server status before it is ready for it and go into an infinite loop
when it finally trys to read the server status.

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
When starting a backup using -i (immediate), and -L (logging) turn on
progress logging for the client, the job doesn't appear in the bpjobd
monitor (Activity Monitor). I get the following error in the
bpjobd debug log: Jobid came as 0.

Currently adding formal ability to use backup progress logging on clients
when using the bpbackup -i, which is currently disallowed by bprd and
bpbackup. Also, note that bpbkar is successfully receiving the -L
option, so bpsched seems to pass along the log file successfully.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
bpsched core dumped on active node of VCS cluster while running jobs.

The problem is: the fixing of memory leaks went a little too far here
and tried to free a literal character pointer.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Check to make sure a groups entry exists in the xinetd
files for the Macintosh 10.1 client. If there isn't one,
remove the existing xinetd file and create a new one.

(NetBackup Clients: MacOSX10.2 )
--------------------------------------------------------------------------------
Description:
The REQUIRED_INTERFACE bp.conf entry has no effect on FreeBSD NetBackup
clients.

(NetBackup Clients: FreeBSD )
--------------------------------------------------------------------------------
Description:
bptm core dumped when the TPC buffer size was set to 256K, and the backup
included a directory with lots of files of less than 512 bytes. The TPC
mover was a Pathlight, which advertises a maximum number of segments per
command of 256, but will support over 4096.

Workaround:
Reducing the TPC buffer size to 128K or less will eliminate the failure.

(All NetBackup Servers: HP RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
The error message displayed when unable to obtain device target information
from the 3pc.conf file was being overwritten with "device not found." This
makes it impossible to detect certain misconfiguration errors.

(All NetBackup Servers: HP RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
The bpsched process core dumped and left a child process still running. The
remove_from_mpxgrp_activelist() function attempts to free pool information,
but the pool pointers are really residue. Changed add_to_active_job_lists
() function to initialize the malloc'd structure.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
If an error occurs that leaves the bpbkar process running on a client
during a Flashbackup, the scheduler will resubmit the backup. This results
in a error 20 because the cache file is still being used by the previous
backup process. A change has been implemented to let bpbrm terminate and
wait bpbkar to finish before reporting the error to bpsched.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
If an error occurs that leaves the bpbkar process running on a client
during a Flashbackup, the scheduler will resubmit the backup. This results
in a error 20 because the cache file is still being used by the previous
backup process. A change has been implemented to let bpbrm terminate and
wait bpbkar to finish before reporting the error to bpsched.

(NetBackup Clients: HP SOLARIS )
--------------------------------------------------------------------------------
Description:
This change includes version 3.23.58 of MySQL for nbdbd.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The bpdgclone command fails when a volume has a VERITAS Volume Manager
(VxVM) Data Change Object (DCO) log.

Errors similar to the following appear in the online_util log:

09:33:22.155 [27646] <2> onlfi_vfms_logf: INF - make_new_vol:
executing /usr/sbin/vxmake -g clone_hdsdg_clone < /var/tmp/DGCLBAAquaGd2
09:33:22.156 [27646] <2> onlfi_vfms_logf: INF - vxvm:vxmake: ERROR:
Dco my_test_dco not in configuration
09:33:22.156 [27646] <2> onlfi_vfms_logf: INF - vxvm:vxmake: ERROR:
getting associations of :
09:33:22.156 [27646] <2> onlfi_vfms_logf: INF - Bad record name
09:33:22.156 [27646] <2> onlfi_vfms_logf: INF - could not vxmake
from input file /var/tmp/DGCLBAAquaGd2

(NetBackup Clients: HP SOLARIS )
--------------------------------------------------------------------------------
Description:
When bperror was run it could not load the localized resource dll
(Japanese), it should have retried using English resource dll's but did
not.

admincmd\bperror Error: Can't find Troubleshooting Guide, or status code
not available.

The resource dll in place: NBWTroubEN.dll

Workaround:
Change the NetBackup Machine Default Locale from default to English.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
During an NDMP restore, and even though the NAS server reported a
NDMP_DATA_HALT_CONNECT_ERROR, NetBackup reports the restore as successful.
(Note: Receiving an NDMP_DATA_HALT_CONNECT_ERROR is rare. Most NAS servers
report an NDMP_DATA_HALT_INTERNAL_ERROR instead.)

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
When doing a Remote NDMP backup, the backup hangs after hitting EOM several
times.

This occurs when the tape hits end of media at the very end of the backup
while Remote NDMP is attempting to flush its internal buffers to tape.

Workaround:
Retry the backup with a different tape. This condition is not likely to
occur a second time on a different tape.

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
If an inline tape copy job starts with multiplexing set to 1, another job
with multiplexing set greater that one will attempt to multiplex with the
first job. The multiplex setting of 1 on the first job caused the job to
run without multiplexing enabled. The second job will fail because it
cannot multiplex with the first job.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
A restore from a Remote NDMP backup that spans tapes may indicate an error
occurred. This happens only with Remote NDMP on a Windows machine.

The bptm log shows "read_data_ndmp returned 0" which indicates the restore
was OK. Following this there is an MTFSF 1 tape command that fails with
IO_ERR.


Workaround:
This error can be ignored. Since read_data_ndmp returned good status, the
restore was successful. It is only the tape positioning that failed.

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
If a restore fails because of a media mount timeout, bprd may core
dump when it finishes processing the restore.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
If the Backup Selection List of an Oracle policy contains a mix of backup
templates and shell scripts, the shell scripts will not run if they follow
a template in the list.

(All NetBackup Windows Clients Alpha Dgux HP Linux2.2 NCR RS6000_433 Sequent4.2 SGI SOLARIS
Solaris_x86_2.6 )
--------------------------------------------------------------------------------
Description:
If bpschedreq -predict is executed with a time that is not the current
day, the exclude dates will not be taken into account.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
An imported backup of a backup copy other than copy 1 would base the
expiration time on the backup time rather than the current time, which
would result in the imported backup expiring too early.


Workaround:
The bpexpdate command may be used to change the expiration of the imported
backup.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When multiple densities are used in the same robot, not all drives will be
used.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Expiring one of the images from a multiple data stream backup causes a
restore to fail even though the images still exist.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Calendar scheduled jobs may not run when administered from a different
time zone.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Block level incremental backups of VxFS 3.5 files using 3 streams will now
complete.

(NetBackup Clients: SOLARIS )
--------------------------------------------------------------------------------
Description:
Fix memory leak during tar restore of migrated files.

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
Restore paths were incorrect for NDMP path-based history.

The list of files for an NDMP backup image are missing the original backup
path. This occurs with path-based file history running NDMP V4.

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
The bpvxver script returns version 1.0 of VxVM on HP-UX, in certain cases.

(NetBackup Clients: HP )
--------------------------------------------------------------------------------
Description:
A 174 error (media manager - system error occurred) occurred when an end-of-
media status was returned while trying to write a end-of-file marker on the
tape. This may occur on the Windows or SGI platforms. The bptm log will
show an entry similar to the following:

00:30:40.803 [3256.4608] <2> io_ioctl: MTWEOF encountered EOM (error 1100),
return EOT

This is followed by an entry indicating "buffers out of sequence", and the
174 error.

(All NetBackup Servers: Intel-NT Sgi)
--------------------------------------------------------------------------------
Description:
The scheduler "predict" is out of sync with what actually gets run.

Additional Notes:
xref 280-067-778

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
bpdgclone command failure with "input line too long"

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
The online_util log is flooded with the unnecessary message "Unable to
obtain mount point", which causes a slowdown in FIM backups.

(NetBackup Clients: HP SOLARIS )
--------------------------------------------------------------------------------
Description:
The UNIX bpbkar would sometimes error while processing a long pathname
with a large number of sub-directories. The problem manifests itself
intermittently, depending on number of sub-directories in the long
patch name, memory on the box, stack size, etc.

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
Drives keep going down every few days. Avrd reports semaphore errors
similar to the following in /var/adm/messages:

messages.0:Oct 28 21:56:28 server1 avrd[8148]: Error in Semaphore
Operation - Result too large
messages.0:Nov 1 03:20:54 server1 avrd[18968]: Error in Semaphore
Operation - I/O error
messages.1:Oct 20 01:31:34 server1 avrd[26744]: Error in Semaphore
Operation - I/O error


Workaround:
Down the NDMP drives that are not working.

(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
Resolves a potential Dr. Watson on Windows NT if an NDMP drive path name is
too long (exceeds the stack).

With this patch, the new maximum length for drive path name is 128
characters.

(All NetBackup Servers)

--------------------------------------------------------------------------------
Description:
If a client is down when the scheduler is doing autodiscovery, and the
Windows GUI attempts to modify a policy, the policy may get deleted.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
NDMP backup restore can fail because some NDMP venders do not use "2" as the
mount point inode number.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Sybase backups fail occasionally with a 143 error if the scheduler is
unable to connect to the client while writing the progress log.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
On UNIX systems, the tldcd and tl8cd robotic daemons may fail if
connections
are opened and closed quickly with no data.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When restoring NAS data from an NDMP-attached tape drive, the bprecover
command fails with an EOM (end of media) error.

Workaround:
Workarounds for this problem:
- Use a non-NDMP drive.



(All NetBackup Servers: HP Intel-NT Linux2.2 RS6000 Solaris)
--------------------------------------------------------------------------------
Description:
A missing tape unmount on one of the copies when using Inline Tape Copy may
cause unnecessary 134 status codes (unable to process request because the
server resources are busy).

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When performing drive cleaning from the java GUI, drive names with more
than 22 characters could cause problems for the GUI.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
A core dump can occur if you run the following command:
bppllist -allpolicies or <policy> -U

Workaround:
To avoid this issue, do the following:
1) Look for a policy that has an unusually long value for one of its
attributes, such as the keyword, or that has many clients.
2) Run 'bppllist <policy name> -U' to display this one policy.
3) If bppllist aborts, modify the policy to reduce the length of
the long value.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Scheduler gets an occasional core dump when it�s preparing to exit. The
error happens when trying to determine which storage units are busy.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Using a Binary Catalog could cause an inablity to browse for restores under
certain file selection lists. This issue has been resolved.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Customers using noauth authentication with the windows GUI would not be
able to start manual backup jobs if they were not an Administrator on the
system running the windows GUI.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Suppose the master server and media server are different servers. Also
suppose that the media server and client are the same machine so that
shared memory backups and restores are normally done.

Suppose the master server is configured so "connect backs" with the media
server are done using VNETD. The typical bp.conf entry on the master server
would look like this:

CONNECT_OPTIONS = mediax 0 1 2

Where "mediax" is the hostname of the media server. The digit "1"
signifiies that vnetd style connect backs are used.

In this case, restores will fail. Typically, the job will fail with error
code 5 and be marked as "incomplete" in the job monitor.



Workaround:
Do not use VNETD style connect backs with the media server if the media
server is also a client.


(All NetBackup Windows Clients All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
Correct an end-case problem in MPX read processing (for
restores and duplications) that may try to unmount a media that is already
in use by another process.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When verifying very large backup images, the Activity Monitor would
incorrectly report a failure status that included a �client process
aborted� error message. Testing showed that the verify job had not
actually failed and the incorrect status was displayed because the
keep-alive checks were not being sent to bpjobd if the bpbrm socket was
active for specified amount of time. This time interval to send
keep-alive packets has been corrected, resolving this issue.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Exchange backups do not support bandwidth limitation throttling.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Backups of large sparse files on SGI could result in bpbkar hanging and
failure of the backup.

Workaround:
None

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
When restoring TIR incremental data, directories could be restored with
incorrect permissions.

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
AIX added a proc filesystem and mount.c needs to be able to identify it in
order for bpbkar to automatically skip it (as it does for other OSes with a
proc filesystem).

Workaround:
Exclude /proc from the backup file list.

(NetBackup Clients: RS6000_433 RS6000_51 )
--------------------------------------------------------------------------------
Description:
If an Oracle restore finishes before all of the data it requires is read,
on the next write NetBackup will get a write error and the restore fails
with a status 5.


(All NetBackup Servers: Alpha HP Linux2.2 RS6000 Sequent4.2 Sgi Solaris)
--------------------------------------------------------------------------------
Description:
bpdbm's host list is not consistent with that of bprd. The current
impact of this issue reported by customer
is "FORCE_RESTORE_MEDIA_SERVER doesn't work for TIR restores."


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Managing calendar-scheduled backups when the master server is in a
different time zone than the Administration GUI causes the jobs to run in
a restricted time window or not at all.

There is no longer a limit on the number of calendar include/exclude dates
within a schedule.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
On a system running Windows 2000 (SP4), you may not be able to browse or
restore the second volume if 'SET HIST=N' is listed in the middle of the
policy's file list.

Workaround:
To avoid this issue, contact your VERITAS customer support office for help
on manually fixing your catalog files.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Duplication of multiple copies using Inline Tape Copy may fail with a 20
error (invalid command parameter). The bptm log (the write side of the
duplication) shows an entry similar to the following:

14:55:48.274 [5586] <16> bptm: not enough parameters specified

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
A media position error can occur when attempting to restore from a copy of
a backup created by Inline Tape Copy. The bptm log will show an entry that
is similar to the following (note the file number of 0):

13:12:27.348 [22187] <2> mpx_read_init: media id A00001, copy 1,
fragment 1 (839680 Kbytes) filenum 0 being considered for restore

The problem occurs when performing multiplexed backups using Inline Tape
Copy, and a new backup is started while waiting for a tape mount due to
hitting end-of-media.

Workaround:
To avoid this issue, the restore may be done from the other copy.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When using Inline Tape Copy, the wrong media-id may be frozen when a media
error occurs on a copy other than the first copy. This happens when the
fail-on-error attribute is set for the failed copy. In this case, the
media-id for the first copy is frozen rather than the media-id for the copy
that obtained the error.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Restoring symbolic links owned by non-root could show an error even though
the symbolic link was successfully restored (giving Partial Success for the
restore job).

(All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
If a schedule is changed from Frequency to Calendar scheduling and
"Last Day" is chosen from "Recurring Days of the Month," bpclsched will
crash. This occurs from both the command line and from the Java GUI.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
VOPIE Enhanced Authentication may not work when host names are longer than
32 characters.
You may see Authentication Failed (error code 160) messages when connecting
to systems with host names or client names greater than 32 characters.



(All NetBackup Windows Clients All NetBackup UNIX Clients)
--------------------------------------------------------------------------------
Description:
bptm fails with a segmentation violation when a media Disk Staging storage
unit sends a duplication job to the master. This issue has been resolved.

(All UNIX NetBackup Servers)
--------------------------------------------------------------------------------
Description:
Exchange restores from multiple images fails with a communications failure
error. The log will have a message similar to the following:

17:58:45 (158515.002) ERR - unable to create object for restore:
Microsoft Information Store:\SG01102\Log files_1074011439 (BEDS
0xFFFFFE30: A communications failure has occurred.


Workaround:
Create a file named /usr/openv/netbackup/NON_MPX_RESTORE to force
sequential restores from multiple images.


(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
When using "bpexpdate -recalculate" or "bpexpdate -backupid" to extend the
expiration date of one or more backups that have multiple copies, the media
expiration may not be correct for copies on removeable media if the first
copy is on disk.

(All NetBackup Servers)
--------------------------------------------------------------------------------
Description:
The W2k platforms were losing MPX groups. The user could not cancel
orphaned jobs within the MPX group, and additional jobs could not be added
to the MPX group.

Windows-specific code was added to display questionable pointer values
prior to releasing the MPX group. Historically, this minor change caused
the problem to "go away." The thinking being that it's a W2k compiler
optimization issue. This fix will either mask the problem, or help isolate
it.

(All NetBackup Servers)
--------------------------------------------------------------------------------



Attachments

NB_45_9_M_279100.sgi.tar (19.4 MBytes)


Legacy ID



279100


Article URL http://www.symantec.com/docs/TECH43741


Terms of use for this information are found in Legal Notices