HP Server Automation Release Notes for the HP-UX, IBM AIX, Red Hat Enterprise Linux, Solaris, SUSE Linux Enterprise Server, VMware, and Windows® Operating Systems

Turn off UAC:

1. In the Control Panel, click User Accounts.
2. In the User Accounts window, click User
   Accounts.
3. In the User Accounts tasks window, click
   Turn User Account Control on or off.
4. If UAC is currently configured in Admin
   Approval Mode, the User Account Control
   message appears. Click Continue.
5. Clear the Use User Account Control (UAC)
   to help protect your computer check box,
   and then click OK.
6. Click Restart Now to apply the change right
   away, or click Restart Later and close the
   User Accounts tasks window. After the
   workaround is performed, any user belonging
   to the Administrators group will be able to
   deploy agents.

Remove all the files from the following three directories,
if they exist, and reinstall the agent.

%SystemDrive%\Program Files\opsware\agent\lcpython15
\Lib\site-packages\win32c om\gen_py\

%TEMP%\gen_py

%SystemDrive%:\Windows\temp\gen_p

Windows servers running a pre-9.0 agent can
perform a scan on an upgraded 9.0 core but not
on a freshly installed 9.0 core because scanning
Windows utilities are retained on the upgraded
core but are absent from the freshly installed core.

Upgrade the agent on the managed servers that you registered
with SA 9.0.

Make sure the gateway can be reached from the managed server.

Agent deployment fails several times due to timeout issues when the uudecode utility
is not installed on the machine to which the Agent is deployed.

Install the uudcode utility on the machine(s) to which Agents are to be deployed to facilitate Agent deployment.

Reprovisioning for Red Hat Enterprise Linux 6 with the ext4 file system is not yet supported.

None.

You can choose to use a Script component to delete stale files
when deploying a Delta Release.

Package Component file import (using the Import
Package feature) is unreliable for files that are larger
than 100MB.

When creating Application Deployment Manager Package Components, large
files may not upload successfully. You may see
an error message similar to:

Failed to upload installer.msi
(sent 238547 of419232 bytes).
IO Error: #2038

Increasing the memory available to the SA Client should
allow for files up to 350MB to be uploaded using the
Application Deployment Manager. Larger files can be
uploaded using the Actions > Import Software menu
when editing a package in the Software Library.
To increase memory available to the SA Client, modify
the “max-heap-size”
parameter in /opt/opsware/occclient/jnlp.tmpl.
For example:

<j2se version="1.6" initial-heap-size="128m"
max-heap-size="1350m"/>

Note that max-heap-size can be adjusted up to ~1350MB
for the 32-bit JVM used by the SA Client.

When you export Application Deployment Manager Environment to a Device Group you receive the following error message:

"java.lang.RuntimeException: Unable to update the device group representing the target '[target name]': DeviceGroup or Server deleted during operation, try again."

To resolve the issue, perform the following steps:

1. Stop Application Deployment Manager
   using the following command:

/etc/init.d/opsware-sas stop da
2. Add the truth.sid configuration parameter
   to the /etc/opt/opsware/da/da.conf file.
   For example:

truth.sid=<newSIDname>
3. Restart Application Deployment Manager service:

/etc/init.d/opsware-sas start da

Perform the following steps after the upgrade :

1. Run the following command to stop the
   Application Deployment Manager:

/etc/init.d/opsware-sas stop da
2. In the /etc/opt/opsware/da/da.conf file,
   remove the truth.sid configuration parameter.
   For example:

truth.sid= <SIDname>
3. Edit the following file:

/opt/opsware/da/webapps/arm/WEB-INF/
classes/hibernate.cfg.xml
4. Locate the block that contains the
   connection.driver and
connection.username properties.
5. Add the following string:
   

   <property name="connection.url">
jdbc:oracle:thin:@ (DESCRIPTION =
(ADDRESS = PROTOCOL = TCP)
(HOST = host1)(PORT = 1521))
(ADDRESS = (PROTOCOL = TCP)
(HOST = host2)(PORT = 1521))
(LOAD_BALANCE = yes) (CONNECT_DATA =
(SERVER = DEDICATED) (SERVICE_NAME =
truth) (FAILOVER_MODE = (TYPE = SELECT)
(METHOD = Preconnect) (RETRIES = 180)
(DELAY = 5))))</property>

   Here, host1 and host1 are variables that represent
   the names of your database servers. Include one
(ADDRESS = (PROTOCOL = TCP)(HOST = hostname)
(PORT = 1521)) specification for every database
   server in your environment.
6. Run the following command to restart the Application
   Deployment Manager:
   
   /etc/init.d/opsware-sas start da
7. Re-run the Opsware Installer—or, alternatively, invoke
   the following command to populate Application Deployment Manager baseline data:
   
   /opt/opsware/da/bin/da_baseline.sh

Deployment into (or undeployment from) a
previously managed but now deactivated server
reports success. It should report a failure.

When a server is deactivated, all Policies are
detached, and all history is discarded.

You cannot add a Tier to an Application if the
Tier name contains certain special characters.

Specifically, the Add Tier dialog will appear
completely empty.

Do not use the following characters in a Tier name:  >
(greater than), < (less than), or & (ampersand)

If you create a Tier using these characters, and you encounter
problems adding tiers, simply delete the problematic new
Tier from Application Deployment Manager
Administration screen.

After upgrading the Java plug-in for your browser
to version 1.6.0_22, the following error dialog
appears whenever the SA Client or Application
Deployment Manager is launched:

"The new Java Plug-In requires a recent version
of the FireFox browser (Firefox 3 or later)"

Click OK to close the error dialog.

Alternatively, you can change the following setting in the
greprefs\all.js file (by default, c:\program files\hp bsa\launcher\gre\greprefs\all.js):

pref("security.enable_java", false);

This prevents the error dialog from being displayed but may
have other consequences. Be sure to restart the SA Client
after you make this change.

In the Google Chrome 10 browser, certain operations in the SA Client (for example: Help > About or Tools > Application Deployment Manager) work once and then become unresponsive. This is due to a known problem with Adobe Flash applications in the Chrome 10 browser.

Chromium project bug: http://code.google.com/p/chromium/issues/detail?id=79591

APX session timeout is not correctly implemented.

To increase timeout for Web APX Proxy (Note: this must be
done at all the Slices on all the Cores):

1. Edit /etc/opt/opsware/apxproxy/apxproxy.conf
   to specify both the session timeout and socket
   timeout. Make sure socket timeout is at least
   equal to session timeout. For example,

   apxproxy.apx.session.timeout=1200000
apxproxy.apx.socket.timeout=1200000

2. Edit both /opt/opsware/apxhttpd/conf/httpd.conf.tmpl
   and /var/opt/opsware/ogfs/mnt/root/opt/opsware/
   apxhttpd/conf/httpd.conf.tmpl to add a timeout directive.
   Add this property after the "Server" property. For example,
   

   imeOut 1200

   Remember, unlike apxproxy.conf, the timeout value in
   httpd.conf.tmpl is in seconds, not millisecon

3. Restart the Core.

In multimaster environments, Application
Deployment data mismatches between SA and
BSA Essention can occurr if the data is replicated
across the Model Repository Multimaster
Component (vault) to where the BSA Essentials
Dataminer is installed within the first few
milliseconds of a minute.

Remove the binary files:

1. From the Navigation pane, select Device ->
   Servers -> All Managed Servers.
2. Right-click the managed server and select
   Run Script -> Select Script.
3. In the Run Server Script window, click Select
   Script to display the Select Script dialog.
4. In either the Browser Scripts tab or the Browse
   Folders tab, select the "post-uninstall script for
   Database scanner for Oracle [unix|win]" script
   and then click Select.
   To view this script, you must have privileges on
   the /Opsware/Tools/Server Modules
com.opsware.server.module.storage.
dbscanner.oracle" folder.
5. In the Run Server Script window, click Next.
6. Click Start Job in the Options pane. Do not specify
   any additional parameters.
7. When the job successfully finishes, the files
   related to the Database Scanner for Oracle
   will no longer be on the cleaned managed server.

The status of an ASM Diskgroup shown in
the Properties view (for one database instance
only) is different than the status shown in the
Database Configuration Assistant (DBCA) view
(a CONNECTED status is displayed instead of a
MOUNTED status).

None.

Note: There is an OEM bug about some tablespaces showing the
incorrect used size. The Database Scanner for Oracle gets the
tablespace used size directly from all of its data files, which
avoid the OEM bug.

Patch rollback of the SA 7.81 patch displays
a failure error:
(Failed to remove software
policy (8710001) ProtocolError:
<ProtocolError for 192.168.161.22/
cogrpc.py: 404 Not found>)

and fails to remove Storage Compliance
(which is a new feature and incompabile with
SA 7.80).

After reporting the error, rollback continues to
the next component. The rollback completes
successfully without other errors and cleans
up all patch-related files and folders on the core.

SA Installer

Install uudecode utility.

SA Installer

In the SA Web Client:

1. Log in as administrator (opsware admin user) and
   navigate to Environment -> Customer -> Opsware ->
   Custom Attributes -> CORD_OPSWwayscripts.
2. Identify the SA core server by checking the value field
   of the custom attribute CORD_OPSWway scripts.
3. Log in to the SA Core server you identified in step 1
   and execute the following commands:
   cd /var/opt/opsware/OPSWpatch/OPSWwayscripts
/scripts./post_after_startup.sh
4. Apply any required hotfixes to the wayscripts.

ISM tool fails to upload ISM into a software
policy because the tool mistakenly detects
the registry key HKLM\Software\
Wow6432Nodeidentified, causing it to
identify a Windows Server 2008 server as a
Windows Server 2008 x64 server.

If you are developing an ISM on a server that has this
registry key, temporarily rename the registry during
ISM development so that ISMtool will correctly identify
the server as a Windows Server 2008 server.

Run the patch script patch_database.sh ensuring
that the file optional_updates.conf is configured as
described in Database Schema Update Procedure
in the Installation section of these release notes.

When we upload a package like this
C:\Program Files\Opsware\ocli\scripts>
oupload -O "Windows 2008" -e gb2312
--pkgtype ZIP -d "/"  c:\download\testpkg中文.zip
we get an error
SUMMARY:
Name: exceptions.UnicodeEncodeError
Description: 'ascii' codec can't encode characters
in position 7-8: ordinal not
in range(128)

RAID deployment fails when the RAID
configuration is captured using ACU
version 8.35.7.0 (linux5 boot image) for valid
RAID configurations on machines with SCSI
drives because the "pretty printing" of SCSI
drive bus values uses 0-based index instead
of 1-based index.

1. Perform captures with ACU Version 8.25.5.0
   (using boot images other than linux5) and deploy
   those.
2. Modify the raid.hpacu.script custom attribute value for
   RAID Array Configuration on ACU Version:
   8.35.7.0-captured RAID policies to use the correct drive
   indexes.
   For example, modify this captured configuration:
   ; Array Specifications
Array= A
; Array Drive Type is Parallel SCSI
; 1:0 (36.4 GB), 1:1 (36.4 GB), 1:2 (36.4 GB),
1:3 (36.4 GB),
1:4 (36.4 GB), 1:5 (36.4 GB)
Drive= 1:0, 1:1, 1:2, 1:3, 1:4, 1:5
to the following:
; Array Specifications
Array= A
; Array Drive Type is Parallel SCSI
; 2:0 (36.4 GB), 2:1 (36.4 GB), 2:2 (36.4 GB),
2:3 (36.4 GB),
2:4 (36.4 GB), 2:5 (36.4 GB)
Drive= 2:0, 2:1, 2:2, 2:3, 2:4, 2:5

After a server is provisioned using a RedHat
DHCP-LESS image in the VLAN, attempting to
reprovision the server causes the server to reboot,
after which the reprovision process fails at the
Anaconda Configure TCP/IP window, prompting
for network information.

When booting a target server using the Linux
boot CD in a DHCP-less network, specifying
an IP address already in use by another host
will prevent the target server from successfully
registering with the SA core. No network
check/warning error is displayed.

After you choose Windows as the OS Family, you
cannot use the Manage Boot Client (MBC) to create
winpexx-ogfs server records in the MBC Single
Form because winpe32-ogfs and winpe64-ogfs
image types are not displayed in the PXE image
drop-down box.

When RAID deployment fails after the RAID
controller configuration has been cleared,
subsequent RAID captures or deployments
will fail unless RAID is first configured manually.

The following error displays:
Exit status: 1280

Error message from ACU: ERROR: (2821)
No controllers detected.

Manually set the RAID controller configuration:

1. Restart the server and press F8 when prompted.
2. Create logical drive(s).
3. Assign physical disks and RAID level.

After the RAID controller has been manually configured,
SA can be used to capture and deploy RAID configurations on
the machine.

Assign Customer OS Build Plan content script fails if
customer names contain quote characters.

In some environments, when writing I18N content to
files from OGFS to WinPE, data and files that contain non-ASCII characters might be invalid.

When you upgrade, unless you have customized the bm.reprovision_attributes_to_preserve system
configuration value, your custom values will not be
replaced by the new values required for this SA release -
instead, old values are retained.

The Cannot access target server over OGFS
error occurs randomly on a server while running an
OS Build Plan on multiple servers.

Run OS Sequence does not escalate device group
privilege. Run an OS Sequence that has an attached
device group results in an exception and the job
is not created.

When provisioning Windows Server 2003 or 2008
using an OS Sequence to a non-C drive using a
WIM image:

• Provisioning works for Windows Server
  2003 OS Media.
• Provisioning fails for Windows Server
  2003 WIM and Windows Server 2008
  OS Media and WIM.
  When provisioning Windows Server 2003
  or 2008 using OS Build Plans to
  non-C drives, Windows 2003 & 2008 OS
  Media installs are supported, but Windows
  2003 & 2008 WIM installs are not supported.
  

Windows Server 2003

When performing OS provisioning using
Application Deployment Automation, if the SA OS
Sequence in use includes a device group and
multiple servers are provisioned simultaneously,
some of those server provisioning jobs will fail with
the following error:
Unexpected general exception: com.opsware.device.DeviceGroupVO.modifiedDate
is set to the illegal value.

Provisioned machines using MBC may continually reboot
into the un-provisioned server pool after initial provisioning
because the sequence_id and MAC link were not removed.

1. Delete the link file (the filename is equal to the target server
   MAC address) from the core's DHCP server's /opt/opsware/boot/tftpboot/pxelinux.cfg directory.
2. If set, remove the sequence_id custom attribute
   from the server.

After migration, Windows Server 2008 R2 x64 servers
are listed incorrectly as Windows Server 2008 x64 in the
SA Web Client, but correctly as Windows Server 2008
R2 x64 in the SA Client.

SA fails to complete OS provisioning if the net or subnet from which the machine obtains its IP address does not have PTR records associated with the IP address.

Provisioning an unmanaged server and bringing it into the Managed Server Pool in a DHCP-less environment (using the linux6 option and HPSA_linux_boot_cd) does not work correctly.

If you have not installed an Agent Gateway on the server hosting the OS Provisioning components (Boot Server, Media Server), running DHCPless provisioning using HPSAlinux_boot_cd.iso results in an error (DHCPOptions.ini file not found).

Red Hat

ESXi DHCP-less provisioning

Follow these instructions:

1. Launch the DHCPless provisioning wizard by starting the HPSAlinux_boot_cd.iso.
2. Choose Static Network Configuration.
3. Enter the OS Provisioning Boot Slice as the Server IP.
4. Enter the Opsware Management Gateway port (3003) instead of the Agent Gateway port.
5. Click Finish.
   

After you have completed these steps, your OGFS-agent will register to the core.

Although the reboot is performed correctly, when you
preview remediating a patch policy on a server, or view
the job status for a patch policy that is already
remediated, the reboot setting might incorrectly
display “Install and Reboot Later” when it should
display “Install and Reboot”.

Importing clusters with old names using
the solpatch_import command will fail because
Oracle has renamed their Solaris patch clusters -
the clusters in your SALibrary may not match the
cluster names from Oracle on Sunsolve.

The solpatch_import –filter option does
not display recommended and/or security
patches if they had previously been marked
obsolete. This became an issue on June 4,
2010 when Oracle changed the criteria for
recommended and security patches
(described here: http://blogs.sun.com/patch/
entry/merging_the_solaris_
recommended_and).

1. You use the solpatch_import –filter option.
2. You have run solpatch_import –update_db on
   June 4, 2010 or later.

1. Log in to the SA Client.
2. Select Library in the Navigation pane.
3. Select By Folder.
4. Navigate to /Opsware/Tools/Solaris Patching.
5. Delete the following files:
   • solpatchdb.zip
   • solpatchdb-old.zip
   • solpatchdb_supplement.zip

If a patch policy contains one or
more superseded patches, the
number of total rules counted after
remediating or patch installation
is incorrect. Note that the
compliance state of the server is
accurate before and after remediating.

Automatically installing recommended
Microsoft patches Q934041 (2000) and
Q924883 (MS07-014) fails. Manually
installing the patches results in a
Non-Compliant server status and a
dialog that is displayed:

Update has already been applied or is
included in an update that has already
been applied.

Windows Patch 944036 (installer for
IE 8) reports install failure (and returns
a non-zero exit code) on Windows Server
2008 x64 managed servers, even though
the patch is installed correctly.
Subsequent patch and compliance
scans will indicate the patch is installed
and compliant.

If you try to install the March 2010
(or later) MBSA patch database, and
then attempt to install MS10-015
(KB977165) and MS10-021 (KB979683)
on to a Windows Server 2008 x86 using
a Windows Patch policy, the Patch install
job results will incorrectly indicate success.
After a patch compliance scan, SA will still
report the patches as recommended for
the server, and the server will be listed
as non-compliant.

1. To install these patches, import the April 2010 (or later)
   version of the BSA Essentials Network patch supplement
   from the BSA Essentials Network on to your SA core server.
2. Visit http://support.microsoft.com/kb/980966/ to download
   KernelSystemStateCheck.exe to determine whether the
   patch can be installed on your Windows Server 2008 (x86)
   managed servers.
3. Contact HP Server Automation Support in order to get this file
   “qc110257.pyc”. (The Quality Center bug report number is
   QCCR1D 110257.)
4. On the SA core, copy qc110257.pyc to the SA core’s Data
   Access Engine (spin) server.
5. As root on the SA Data Access Engine (spin) server, execute
   the following command:

   # /opt/opsware/bin/python2 qc110257.pyc

6. Assuming Microsoft's KernelSystemStateCheck.exe utility reports a pass result, use the ad-hoc Install Patch task window to install the version of MS10-015 (KB977165) that has a file name of
   Windows6.0-KB977165-x86.msu, and the version of MS10-021
   (KB979683) that has a file name of Windows
   6.0-KB979683-x86.msu.

Attempts to install the patch for
KB944036 (Installer for IE8) using SA will fail.

Windows Server 2008 x86

After remediating the vendor recommended
policy on a Windows managed server,
the server's recommended patch list shows
additional patches that need to be installed,
requiring the vendor recommended policy
to be remediated again. This sometimes
occurs when older patches introduce
problems that newer patches already fixed.

105671

User’s  preferred  datetime format is:
“星期五 十月 22, 2010 15:16 CST”, but in
the SA Web client, all dates and times are
displayed as “Fri Oct 22,2010 14:04 CST”

Chinese Date format: Commonly used date
formats, like yyyy ’year’ mm ‘month’ dd ‘date’
and ‘yyyy-mm-dd’ are not listed in the user
profile setting, so users cannot select a
favorite date format.

SA 7.8x and SA 9.x with NA7.5x integration
fails.

Perform the following actions:

1. In the jboss_wrapper.conf file, comment out the
   following lines: #wrapper.java.additional.6=-Dorg.omg.
   CORBA.ORBClass=com.sun.corba.se.internal
   .Interceptors.PIORB
   
   #wrapper.java.additional.7=-
   Dorg.omg.CORBA.ORBSingletonClass=
   com.sun.corba.se.internal.corba.ORBSingleton
   
   #wrapper.java.additional.8=-
   Xbootclasspath/p:/opt/NA/server/ext/wrapper/lib/
   CORBA_1.4.2_13.jar
   
   
2. Restart truecontrol.

The SA/OO Integration feature is not
available until after you import the required
OO SDK Client Certificate.

Note: OO SDK Client certificates support
different versions of OO. SA 9.01 contains
an OO SDK Client Certificate for OO versions
7.51, 7.6, and 9.0. Make sure you have a
certificate that is compatible with
your SA version.

1. Enter the following command to stop the twist
   component:

   /etc/init.d/opsware-sas stop twist

2. Enter the following command to export the OO
   Central Certificate:

   /opt/opsware/jdk1.6/jre/bin/keytool
-exportcert -alias pascert -file
/tmp/oocentral.crt -keystore
/var/opt/opsware/twist/oocert

3. Import the OO Central Certificate to the SA JRE Keystore.
   The JRE keystore default password is changeit.

   /opt/opsware/jdk1.6/jre/bin/keytool
-importcert -alias pas -file
/tmp/oocentral.crt -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts

4. Make sure there are no errors when you entered the
   previous commands.
5. Enter the following command to make sure the
   certificate was imported successfully:

   /opt/opsware/jdk1.6/jre/bin/keytool
–list –alias pas –keystore
/opt/opsware/jdk1.6/jre/lib/
security/cacerts
   
   Example output:
pas, Feb 3, 2010, trustedCertEntry,
Certificate fingerprint (MD5):
DF:DD:22:1B:A2:1E:A9:9C:1C:AF:8F:
E0:14:1F:B5:E0

   Enter the following command to start the twist:
/etc/init.d/opsware-sas start twist

OO servers use a certificate that is bundled with SA 9.10 and that is used to connect SA to OO. This certificate expires May 21, 2011. As of today, a newer OO certificate is not available to bundle in SA.

The certificate will be updated when you apply the next OO patch, but, as a consequence, the SA OO integration will fail. If this happens, you must manually import the certificate into SA.

NOTE: If you do not update OO, the integration will continue to work.

To manually install the certificate:

1. Stop the Web Services Data Access Engine.
2. Transfer the OO certificate to SA:
   1. /opt/opsware/jdk1.6/jre/bin/keytool -exportcert -alias pascert -file /tmp/oocentral.crt -keystore /var/opt/opsware/twist/oocert
   2. /opt/opsware/jdk1.6/jre/bin/keytool -importcert -alias pas -file /tmp/oocentral.crt -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts
3. Check that the OO Central Certificate was imported successfully:
   1. /opt/opsware/jdk1.6/jre/bin/keytool -list -alias pas -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts
4. Start the Web Services Data Access Engine.

When a new OO certificate becomes available, use the following process to install it:

1. Stop the Web Services Data Access Engine.
2. Transfer the OO certificate to SA:
   1. Manually obtain the new OO certificate.
   2. Save it in: /tmp/oocentral.crt
   3. Enter the following command:

      /opt/opsware/jdk1.6/jre/bin/keytool -importcert -alias oocentral -file /tmp/oocentral.crt -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts

3. Check that the OO Central Certificate was imported successfully:
   /opt/opsware/jdk1.6/jre/bin/keytool -list -alias oocentral -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts
4. Start the Web Services Data Access Engine.

If you follow this second process, the new and old certificates can coexist on the server. Co-existence might be useful if SA integration with an older OO is required.

If you want to use the same alias name (pas alias), you must delete the older certificate first:

/opt/opsware/jdk1.6/jre/bin/keytool -delete -alias pas -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts
/opt/opsware/jdk1.6/jre/bin/keytool -importcert -alias pas -file /tmp/oocentral.crt -keystore /opt/opsware/jdk1.6/jre/lib/security/cacerts

In the SA Client in an NA-enabeld core,
if you run the Connections by Network
Device report and set the parameter to
Device Name Equals [Any Value], the search
returns no results.

The “Application Deployment” and “Patch”
folders and their reports in the BSA
Essentials/SAR Client 2.0/2.01 installation
patch are missing. These folder definitions
are required to view the Application
Deployment and Windows patch reports.

Log in to a BSA Essentials core server.

1. Change your user to become the BSA Essentials
   super user on the server:
   su - ombd
   
2. Enter the following commands:

   
   cd /opt/opsware/omdb/deploy/birt.war
cp report_def.xml report_def_org.xml
vi report_def.xml

   Add the following two lines in the section of the
report_def.xml file:
   
   

   application_deployment_reports

   top_level_patch_reports

   Example: <folder name="reports">

   <display_name>Reports</display_name>

   <folder_reports></folder_reports>

   <subfolders>

   <subfolder_name>
application_deployment_reports< /
subfolder_name>

   <subfolder_name>top_level_patch_reports
</subfolder_name>

   .....

   .....

   /subfolders>

   <parent>none< /parent>

   < /folder>

   Save the report_def.xml file.

   Launch the BSA Essentials Java Client.

   Click on the Reports folder to see the
   “Application Deployment” folder and
   “Patch” folders.

Internet Explorer enhanced with Security
Configuration (IE ESC) that is enabled to
access the SA Web Client blocks some features
of SA Web, such as the search function.

In some cases, in the SA Web Client Manage
Gateway page, changes to bandwidth throttling
values between two gateways do not persist.

You must apply any bandwidth changes between
gateways on the Core
Gateways only. Changes made on other gateways
will not take effect.

Network scans to a satellite realm fail for
hosts with the error:
XML document structures must start
and end within the same entity.

Exporting a Run Server Script Job output
containing multi-byte characters
(Japanese/Korean) to .csv results in question
marks in place of the multi-byte characters.

There is no Target and Target Volume
information displayed in the storage volume
access path view for a LUN

When you perform a provisioning operation
for an HP EVA array (such as create, delete,
or modify a volume or pool), the changes for
the volume or pool might not be immediately
available in the corresponding Inventory view
in SA after running the "Update from Storage
Essentials" process.

When a server on which SE Connector is
running is directly deactivated and deleted,
selecting Administration -> Storage Scanners ->
<The names of the SE Scanner on "Host">
shows stale entries of Storage scanners in the
Scanner panel. The stale entries count will
increase, depending on how many times the
server is deactivated and deleted from the core.

Restoring OS Provisioning Stage 2 images
fails on SUSE Enterprise Linux 9, which is a
deprecated platform.

You cannot set the timeout value for the time
it takes to install or remove software or execute
scripts to anything other than the default value
of 5 hours - the job times out at 5 hours.
This timeout value is specified by
"way.remediate.action_timeout" in the
SA Web Client.

The job fails with the message "The request
to retrieve information from the Agent failed
because it timed out. If the problem persists,
please contact your HP Server Automation
Administrator."

After performing a software remediation, the
compliance status might incorrectly show servers
as being out of compliance, because there is a
caching delay in the Web Services Data Access
Engine (twist).

Software Compliance scan status displays a Scan
Failed status after attaching and remediating a
software policy if the file solpatchdb.zip (the solaris
metadata database) is missing.

If you run a large remediate job (for example,
500 servers), you might encounter a Web
Services Data Access Engine (twist) cache
full exception while running compliance
across the servers.

When using WebService API to invoke
SoftwarePolicyService.create(), if you set the
RPM package in the installable item list by using setInstallableItemData(), the install list is
not created.

Migrating a Windows Server 2008 R2 x64 server,
attached to a dynamic device group with the rule
Operating System=Windows Server 2008 x64
specified, prevents all Application Configurations,
remediated using a Software Policy, from being
detached from the server.

LUNs that are multipathed by EMC PowerPath
are shown as both "ROOT" and "LUN" in the
Inventory -> Storage -> Volumes Panel
alongside LUNs that are correctly displayed.
The SCSI Bus number provided by PowerPath
(using the powermt command) does not match
the bus number of the disks (LUNs).

The Server -> Relationships -> SAN Switches
panel on a virtual server is empty/only displays
SAN switches to which the given server is
directly connected.
In some cases, a server may depend on SAN
switches that are not displayed in this panel.
For example, a virtual server may be using
storage allocated from a hypervisor that was
allocated storage from a SAN.

When you run the storage snapshot specification,
the disk capacity information is changed in the
Inventory -> Storage -> Disk panel. This occurs
if there is a mismatch in the disk names, as
reported by the hardware registration script
and the storage snapshot specification.

An EMC Symmetrix array that is discovered
through SE Connector can report more than
one storage volume with the same LUN number
presented to a managed server. Running the
storage snapshot specification on the managed
server will succeed; however, the Inventory
-> Storage -> File Systems and Inventory
-> Storage -> Managed Software panels
will be empty. In addition, some host volumes
with a LUN service type will not be displayed
in the Storage -> Inventory -> Volumes panel.

The right-hand panel under
Inventory -> Storage -> Volumes does not
display the supply chain and composition
information properly (or the information is
missing altogether) for some volumes that
were created out of LUNs

For managed servers with mirrored volumes,
if one of the disks that is part of a mirrored
volume fails or is removed, the state of the
volume is shown as "Failed Redundancy"
in the Disks Management panel. However,
in the Inventory -> Storage -> Volumes panel
for the managed server, the status of this
volume is shown as "OK", even after running
storage snapshot specifications.

Running a storage snapshot specification on
unsupported ESX 3.0.x servers returns an error
message that indicates unsupported namespace
in content of SOAP body. Storage related
information for ESX 3.0.x is not stored and displayed
in the SA Client.

Host Storage Inventory does not support VMware
servers that have VMDK on NFS datastore. Therefore,
no storage information will be collected on VMWare
servers with this configuration.

Solaris LVM RAID on Soft Partition on slices stops
responding, resulting in a defective storage supply
chain.

The SNIA v1 HBAAPI Extension reports two single
port cards when a single dual port card is present
in a multi-port card.

When you add a mirror to concatenated or stripe,
there is no distinction between the volume types
"Mirror Concatenated" and "Mirror Striped" in the
Volume Manager labels. Note that
"Mirrored Concatenated" and "Mirror Striped" are
distinct on the volume manager on the host, such
as on the Veritas Volume Manager.

The value stored in
STORAGE_COMPONENTS.STORAGE_DRIVE is a
different format on Solaris 5.10 than on Solaris 5.8
and 5.9. The different format for 5.10 causes a
broken storage supply chain on affected servers.

On a Windows 2003 server with the SNIA library
from QLogic 9.1.4.15 HBAAPI, Fibre Channel
Adapter and storage volume information might
not be discovered by the Storage Host Agent
Extension, causing fibreproxy.exe to stop
responding.

When you snapshot a Unix server that has a
QLogic driver installed, there is no:
FC adapter information in the Hardware view.
Composition and connectivity information for any
SAN volume in the Volumes pane.

Veritas Storage Foundation 4.3 with QLogic
9.1.4.15 results in invalid fibre proxy SCSI
addresses. The SAN storage volume displays
both LUN and Root as the Service Type. There
are two lines for the physical drives: One line
displays LUN and the other displays Root.

None

Note: The Storage Host Agent Extension does not report file
systems that have non-drive letter mount points.
The Storage Host Agent Extension does not report file systems
that have multiple mount points.

Fibreproxy is broken on Windows 2000 SP4 server
with a QLA2310 HBA and vendor driver version
9.1.4.10. A storage inventory snapshot does not
gather and supply complete data, including storage
volume and FCA information.

When you run take a Storage Host Agent Extension
snapshot by running fibreproxy on a Windows server
where Emulex LP850, LP952, LP9002, or LP9402 is
installed, three FibreChannelTargetMappings are
returned, two of which are duplicates. This symptom
does not occur with Emulex driver 1.30a9.

If you run the chpath command as shown below to
take a Storage Host Agent Extension snapshot for
each available path to the device, all the MPIO paths
to a logical device become disabled. In this state,
the system calls used by the diskproxy and mpioproxy
will stop responding.
chpath -l hdisk2 -p fscsi0 -s disable xx

After you update a LUN mapping (and remove the
initiator ports) from a mounted SAN volume on a
server, the old LUN mapping information still
displays in the SAN array volume view and in the
server storage volume view. An additional access
path is displayed in the SAN array volume view
(Access Path subview) for the volume for which
LUN mapping was removed. The access path
that shows no initiator device and/or initiator port
information is the correct one. The incorrect access
path is removed from the display when the next
Storage Host Agent Extension snapshot is run.

The SAV displays incorrect information after adding
a zone to a fabric. The zone is not associated to the
correct server/port/WWN.

The multipath information is not reported correctly
for a server that has HP-UX 11iv2 OS installed and
Veritas DMP managing the multipathing in the
SA Client. The SNIA library does not support HBA_GetFcpTargetMappingsV2r.

On servers running AIX 5.2 with PCI-X Fibre
Channel Adapters, the supply chain does not
display after taking an inventory snapshot.

If you disable a volume in Veritas DMP and
subsequently take a new Storage Host Agent
Extension snapshot, it appears as if the the
updated volume is still managed by Veritas DMP.

After the SA 7.5 release, one of the VMM library jar files
was changed from OPSWvmm-vmware.jar to
vmm-vmware.jar. When an upgrade is performed from
SA 7.5 directly to SA 9.0, or from SA 7.5 to SA 7.8 to SA 9.0,
and any virtualization operation is invoked on the ESX
hypervisors, the VMM package gets remediated first with
the new package but the OPSWvmm-vmware.jar file is
left untouched. This causes the consecutive virtual
machine create operations to fail.
The following workaround is intended for any ESX hypervisor
that is managed by an Agent in SA and whose virtualization
aspect will still be handled through the same route (not via VS)
in SA 9.0.

1. Create a script.
   1. From the Navigation pane, select Library -> By Type -> Scripts ->
      Unix.
   2. From the Actions menu, select New and then enter the following
      information:
      Name: Clean VMM on ESX
      Location: Select Package Repository -> All VMWare Linux ->
      VMWARE ESX Server <any version>
      Script Content:
      1. unlink /opt/opsware/vmm/lib/OPSWvmm-vmware.jar >
      /dev/null 2>&1
      
   3. Description: Enter a brief description, as necessary.
   4. Leave the defaults for the remaining fields and then select
      File -> Save to save and close the window.<
2. Create a dynamic device group and add the target servers
   to this device group.
3. Run the script on the servers in the device group created above.

Note: This script can be used on all ESX versions.
There is no need to duplicate it in different packages
or to create a separate device group for each hypervisor version.

The Web Services Data Access Engine
does not start and records the error:

Could not reserve enough space for object
heap in:
/var/log/opsware/twist/boot.log

The default maximum JVM heap size has
been increased to 2560 MB, and as a result,
the Web Services Data Access Engine (twist)
does not start properly on Linux AS3 32-bit
systems due to a two-gigabyte memory limit
for a single process running on JVM on 32-bit
systems.

Before upgrading, edit the file:

/etc/opt/opsware/twist/twistOverrides.conf
and add the following entry:
twist.mxMem=<memory size in Megabytes>
The value must be 2000 megabytes or less.
Example: twist.mxMem=1960m

Cannot specify multiple LDAP servers for redundancy.

Support needed for external authentication redundancy for LDAP servers in twistOverrides.conf file.

Discrepancy between cpu_family reported by different Agents on the same hardware.

When exporting an Application Deployment Manager environment to a device group with targets that contain servers that have been deleted, you received the following error message:

“java.lang.RuntimeException: Unable to update the device group representing the target '[target name]': DeviceGroup or Server deleted during operation, try again.”

“Server and policy platform mismatch” error occurs at deployment time when using tier policy
If the server platform did not match the platform of the policy associated with the tier, a deployment job was started, but it immediately failed. This was inconvenient, since the mismatch should have been reported earlier.

If you cancelled a job that just started, the confirmation dialog contained a message that said:  “Are you sure you want to cancel job?”

Backup and rollback scripts for large Code components were failing, because the default timeout for these scripts was very short.

The Windows Backup and Rollback scripts were not handling directory hierarchies correctly
For full releases, the Windows Backup script backed up files into multiple places and created too many files when the Source Directory had subdirectories. For delta releases, it did not back up files in subdirectories correctly.

Windows backup and rollback scripts don't work properly if drive letter is not specified for the Default Install Path.

Parameters grid on the Deployment screen should show components like the Component Editor does so that you can see where parameters come from.

Parameter values were incorrect if an application included more than one OO Flow component.

If two or more environments contain targets with identical names, the incorrect environment is sometimes shown in the Parameter Editor dialog.

Application Deployment Manager should warn users in the Job Step output when a Script component times out. There is no indication in the Output area of the job Step details window that a timeout had occurred. This is the window that is displayed when you double-click a job step on the Jobs screen (or in the Debugger) or when you drill-down from the Job details window.

( The Output area in the job Step details window now indicates the reason that a Script component fails.)

Any SA job can be blocked pending approval. Application Deployment Manager deployment jobs are subject to this blocking. Previously, if an Application Deployment Manager deployment job failed because it was blocked pending approval, no message to that effect was displayed.

When you used a special variable in a parameter to refer to a server, any output that contains the server name would show the SA name for that server, not the host name.

You can add deactivated servers to a target, but you cannot deploy to them.

The Filter box in the Select Targets dialog enables you to show only those servers whose platform matches the selected version. Previously, if you selected the Filter box, and there were no valid servers to choose from, you would see a cryptic error message.

tar options used by Application Deployment Manager are not valid on all platforms.

Backup step on Solaris platforms incorrectly reported “nothing to back up” when a backup was, in fact, performed.

Code component changes list are sorted.

Long Application Configuration names do not display correctly in Application Deployment Manager.

Previously, you could add the same user or group to the list on the Permissions tab for applications and environments. Users and groups that have already been added to the list are removed from the drop-down list of available choices.

tar options used by Application Deployment Manager are not valid on all platforms.

Backup step on Solaris platforms incorrectly reported “nothing to back up” when a backup was, in fact, performed.

Code component changes list are sorted.

Long Application Configuration names do not display correctly in Application Deployment Manager.

Previously, you could add the same user or group to the list on the Permissions tab for applications and environments. Users and groups that have already been added to the list are removed from the drop-down list of available choices.

In the Edit Application and Edit Environment dialogs, there was no distinction between an individual user and a user group in the table on the Permissions tab. There is now a Type column that says User or Group

Previously, the Add Servers to Target dialog listed servers running ESX and servers in the “Unknown” OS category when you were adding servers to a Windows tier. Only Windows OS servers are listed for Windows tiers; only UNIX servers are listed for UNIX tiers.

Previously, if a job failed because the registry key is invalid, the job did not complete with failure status.
All registry key names and values are now carefully validated when components are saved.

Searching for an OO Flow with no characters in the Search box fails.
Previously, if you clicked the Search button in the Select OO Flow dialog, and you had not entered anything in the text box, the search would time out without returning any items.

Previously, binary registry key values in the Add Value dialog box were displayed as hexadecimal.

Previously, if you initiated roll back or undeploy of a version that was older than the most recently deployed version of the same application, you were not warned about or prevented from doing that. This is dangerous and strongly discouraged.
Now, you are warned if you attempt to roll back or undeploy an older version. You must confirm your intention to proceed.

 In a Windows Registry component, if you specified the value of a DWORD type key using decimal characters, the key would not be created. Only hexadecimal characters worked.

If Application Deployment Manager is stopped, any deployment jobs that were running at the time never complete and appear to be IN_PROGRESS in perpetuity.

Need horizontal scroll bar in “Add Servers to Target” dialog.
When you opened a hierarchy of Device Groups—such as Public : Opsware : Operating System : All Linux : All Linux Red Hat—you could not see the full names of the child device groups, because they were truncated and there is no horizontal scroll bar

“Failure parsing template” error reported after an Application Configuration is selected.
When you selected an Application Configuration, if Application Deployment Manager failed to parse the template used with the Application Configuration, and cryptic error messages were reported.

Should not be able to select the same Package more than once.

Incorrect Job ID referenced in confirmation message for rollback or undeploy
When trying to roll back or undeploy a version with OO Flow or Configuration File components, the following confirmation message is displayed:

Say that you scheduled a deployment job to stage immediately but cut over at a different time. Later, you changed your mind and modified the scheduled cut over time to be on a different date.

Previously, if you deleted a release, you could not ever use its name again. You can now recycle release names.

ADM experiences memory problems when there are many Code component changes to show
When creating a new version of a full release, if you selected “Show Code Component Changes,” and there was a very large number of files included in the Code components, you might see cryptic “out of memory” error messages.

When you create or modify a lifecycle, the list of Available Environments is not sorted, causing problems if numerous environments are listed.

(The list is now sorted alphabetically by name.)

The following misleading message appears in a pop-up dialog box after the job started when the“Show Jobs screen when deployment starts” option is not selected:

“One or more deployments have been scheduled. View results in the Jobs tab.”

This message is incorrect. At this point, the job is actually running and is no longer in the Scheduled state.

(The message now reads: “One or more deployments have been started.”)

RPM and MSI packages were imported as type ZIP. The name of the “Import Package” tool used when specifying a Package component is confusing, particularly in the following scenario:

• If an RPM or MSI file is added to a Package component using the the “Add Package” tool, that RPM or MSI file is copied to the “Default Installation Path” on the target server during deployment.
• If an RPM or MSI file is added to a Package component using the “Import Package” tool, however, the ZIP file containing the RPM or MSI file is copied to the “Default Installation Path” on the target machine during deployment.

(The name of the button was changed to “Create ZIP Package from Files.”)

On the Deployment screen, you should not be able to select a Target in an Environment that is not part of the Lifecycle associated with the specified Version.

( Now, only those Targets that belong to Environments that are part of the Lifecycle associated with the specified Version are available.)

Symptom: Occasionally, the following refresh error message is displayed when the Step details dialog is open while a job is running, and the job failed:
“A data refresh failure has occurred. No further refreshes will occur.”

(The Step details window is now dynamically refreshed to reflect any status changes that occur until the job runs to completion.)

When a non-superuser with View All Jobs permission, but no Create Applications or Manage Application Deployment permission, views a job that they do not own, the Job Steps and Parameters tables were empty.

(Both the Job Steps and Parameters tables now show the correct information for the job.)

Automatic backup steps for some components are failing due to timeout errors.

( The timeout for backup steps was increased from 5 to 60 minutes.)

In the case where an Environment contained some targets that were suitable for the specified Version and some that were not suitable, the Environment name was “grayed out” in the Select Targets dialog in the Deployment screen. This made it appear that no Targets in that Environment were available when, in fact, some were available.

(All Environments that contain suitable targets are now shown as available.)

When you create a new deployment on the Deployment screen, the Stage selection should be “Now,” and the Cut Over time should be “Immediately” by default. Instead, the values from the previous deployment are selected - they are not reset.

(The correct default values are now selected.)

Scripts were not consistently updated during a Patch Upgrade. Some “out of the box” backup and rollback scripts were not properly updated after an SA patch was installed.

(All scripts are now updated when a patch is applied.)

APX session timeout is not correctly implemented.

To increase timeout for Web APX Proxy (Note: this must be done at all the Slices on all the Cores):

1. Edit /etc/opt/opsware/apxproxy/apxproxy.conf to specify both the session timeout and socket timeout. Make sure socket timeout is at least equal to session timeout. For example,

apxproxy.apx.session.timeout=1200000
apxproxy.apx.socket.timeout=1200000

2) Edit both /opt/opsware/apxhttpd/conf/httpd.conf.tmpl and /var/opt/opsware/ogfs/mnt/root/opt/opsware/apxhttpd/conf/httpd.conf.tmpl to add a timeout directive. Add this property after the "Server" property. For example,

TimeOut 1200

Remember, unlike apxproxy.conf, the timeout value in httpd.conf.tmpl is in seconds, not milliseconds.

3) Restart the Core.

If the Command Engine is too busy or otherwise unavailable, PENDING jobs cannot be run, however, when the Command Engine is available again, the missed PENDING jobs run, regardless of how old they might be.

PENDING jobs now have a default limit of 600 seconds after which they are considered stale jobs and will not be run when the Command Engine becomes available again. (This limit is configurable but you must contact your SA Support Representative to change the default time.) For example, you have a one-time job that is scheduled to run at a future date and time. When it is time for the job to run, the Command Engine is down. When the Command Engine comes back up, the PENDING jobs are inspected, to check whether or not the job should run based on when the job was scheduled versus the current time, and if too much time has passed (600 seconds by default), the job is not run.

Independent

The get_firmware.sh script does not apply to SuSE 10 SP3 PPC64.

The Extensible Discovery get_firmware.sh script does not support the PPC64 platform.

Oracle RDBMS 11.1.0.7
Database user TRUTH statistics collection job fails with error: ORA-01000: maximum open cursors exceeded.

When duplicating a zip package, the package is duplicated without the path and the following error message is displayed:

You do not have permission to view details of this policy item.

Import Media script should support long filenames for SMB.

The version of the Samba server installed on the remote host during Linux OS provisioning has problems with securing the environment and should be upgraded to a secure release.

In a core where the value for truth.dcNm is not the same as the value for truth.dcDispNm, or truth.dcDispNm is not specified in uppercase only, Linux reprovisioning becomes interactive and prompts the user for the language, locale, etc.

The autoprovision for registerManagedNodeForDS would always send kickstart information through eth0.

If you attempt to specify the Windows Server 2008 R2 volume license product key information by using the SA ProductKey custom attribute for Windows 2008 R2 OS Provisioning, the Windows setup process fails with the following error

Run action is enabled in the OS Build Plan window for a user who does not have the “Allow Execute OS Build Plan” privilege.

The configuration value: bm.reprovision_attributes_to_preserve should default to change to a new value, rather than keep the old value.

You can upload an unattend.xml file that exceeds the maximum size allowed.

Panel title for the Inventory -> Disks view displays incorrect title - it displays the server name.

Improper display of multibyte characters when using IIS and COM+ on Korean servers

IIS - Inventory fails on Italian/German/Korean/Japanese managed server using IIS SM.

Remediation of Test sites does not copy all values from the source to the target.

Environment verbs should not be evaluated.

IIS7 Server Module audit and remediation issue with several applications in the application pool.

During Preview/Remediate, when there is insufficient disk space under /var directory, the following error should display:
insufficient disk space
However, this error is actually displayed:
The request to retrieve information from the Opsware Agent failed for an unknown reason, please contact your Opsware Administrator.Execution error:
Traceback (innermost last):
File "./base/wayfuncs.py", line 133, in evaluator
File "<string>", line 1212, in ?
File "<string>", line 1174, in main
File "<string>", line 586, in installPCA
File "<string>", line 382, in findTmpdir
AttributeError: debug

102109

When attempting to remediate a Linux server, the operation fails with an error such as:
The remediate operation cannot be performed because an error occurred during preview.
/var/opt/opsware/yum/cache/opsware/repomd.xml:9: parser error

When an HPUX 11.23 server is presented with a LUN via dual port HBA, the user interface shows only one FC target mapping.

For LUNs managed by Solaris Native MPXIO, the multipath count is displayed as 1. This behavior is seen on Solaris versions before Solaris 5.10 Update 3.

When an imported LDAP user clicks the Next button in the Create Virtual Zone wizard’s Zone Definition step, they do not advance to the next step. The following NullPointerException error is returned:

WARNING init(NguiUser:-1): No user date format specified, using JVM defaults
java.lang.NullPointerException
at com.opsware.ngui.virtual.task.ZoneDataSpecificationPanel$
ServerVirtualizationZonePanel.createPanel(Unknown Source)
......
at java.awt.LightweightDispatcher.dispatchEvent(Container.java:3128)

When more than 2 GB memory is allocated for Local Zones, requests for memory information causes Web Services Data Access Engine errors.

Update ESX VMM code to version 2.5.

A completed Create Virtual Zone job does not show the zone creation command script.

ESXi hypervisor in maintenance mode: You cannot create virtual machines when it is not allowed.

The reported OS property for ESX servers is inconsistent between direct (SA Agent) and indirect managed (vCenter) cases.

Cannot select Windows 2008 R2 as VMware Guest OS if you create a virtual machine on ESX Server and provision Windows 2008 R2.

60634

In certain cases, for example, moving a twistOverrides.conf file from one core to another that has different encrypted key material, the operation fails and the twistOverrides.conf file is truncated.

Resolution: Changes to changes to the twistOverrides.conf file are now written to a temporary file and the original is only overwritten on successful completion of an operation.