test

Sun Connection 1.1 Release Notes

Chapter 2 Known Issues and Workarounds

This chapter includes information about issues with the Sun Update Connection – Enterprise product and workarounds for these issues.

Known issues are organized as follows:

For a list of fixed issues, see Chapter 3, Issues Fixed and New Functionality.

Installation Issues

Upgrading To Version 1.1.1 Will Reset The uce.rc Files To The Default Settings (CR 6568370)

When upgrading to version 1.1.1, the parameters in the server/cgi-bin/.uce.rc file are reset to the default settings. If you customized any parameters, you must reenter them in the configuration file.

Note –

This does not affect the engine/bin/.uce.rc files and agent .uce.rc files.

Installer Might Fail on 64-bit S/390 Platforms (CR 6517521)

Description:

The Sun Connection 1.1 installer may fail to start on S/390X (64-bit) platforms.

Cause:

The installer uses an application which was compiled for S/390 32-bit and will crash if 32-bit libraries on which it depends are not present. Although this issue has only been seen on RedHat distros, it may also happen on SuSE 9.

Workaround:

Install 32-bit versions of these packages:

  • ncurses-5.4-13.s390.rpm

  • zlib-1.2.1.2-1.s390.rpm

Installer Might Halt With Progress Bar Showing 0% On All Platforms

Description:

During an installation of a Sun Connection 1.1 component, the installation may halt with the progress bar showing 0%.

Cause:

If there is a failure during installation, the installer tries to display a message. However, due to a feature of the application used to display installation progress, the error message is blocked and instead, the progress bar is set to 0% and the installation is paused.

Workaround:

Press Return to continue, exit the installer and consult the installation logs to determine the cause of the failure during installation.

Agent Issues

Fail to Register All NCOs (CR 6565602)

Description:

A bug in the sed utility ( /usr/bin/sed ) during the pkg_seeker.sh process causes failures in the registration of the packages.

Workaround

Install patch #110165-05 or higher.

The patchadd -G Command is Used Incorrectly (CR 6533814)

Description:

This error occurs in Sun Connection 1.1.0 and 1.1.1.

When the agent is installing patches on Solaris 10 in the global zone, it only uses the SUNW_PKG_ALLZONES package parameter to decide whether to use the patchadd command, or the patchadd -G command.

The /var/sadm/install/gz-only-packages file contains a list of packages that were installed on the systems using the pkgadd -G command and that should not be propagated to any local zone. Sun Connection should check the /var/sadm/install/gz-only-packages file to decide which patchadd command to use.

If a package is not in the package list in the /var/sadm/install/gz-only-packages file, it should be installed with patchadd command, not the patchadd -G command. The current agent design incorrectly uses only the patchadd -G command to update the /var/sadm/install/gz-only-packages file.

Results:

Once the /var/sadm/install/gz-only-packages file is updated with the patchadd -G, you cannot properly create a new zone because not all of the listed packages will be propagated into the new zone. This will result in a broken zone,

The patchadd Command Requires You to Manually Mount Zone Roots For Single User Patches (CR 6548606)

Description:

For patches requiring application in single user mode, that alter ALLZONES packages, the patchadd is unable to apply the patch if zone roots are not on the / or /var filesystems.

Workaround:

Before you run a patch job, run the mountall -l command to ensure local filesystems are mounted.

Local Zone Patching With Mounted Partitions

Description:

If a zone mount command fails, patches appear to be installed in multi-user mode. When the zone mount fails, the agent script exits and the system comes up in multi-user mode. The agent restarts and attempts to run the rest of the job. If patches are applied to all zones using the patchadd command, the patches are not added. If patches are applied using the patchadd -G command, the patches are added, but the system does not reboot.

Workaround:

Before you run a patch job, run the mountall -l command to ensure local filesystems are mounted.

Edit the /etc/rcS.d/S90agent file and remove exit 1 from line 130.

Read-Only Users Can Deploy Jobs (CR 6553436)

Description:

In the 1.1.0 and 1.1.1 releases, read-only users can deploy jobs from the send job option in the Report windows. The deploy button is not disabled in the Report output pages.

Agent Installation Might Fail When Logged Into a Zone Using zlogin (CR 6513395)

Description:

When trying to install the agent package in a non-global zone, the user may see the following and the installation will fail:

./Install

/uce_agent/scripts/agent_main: test: unknown operator 80

Cause:

The installer needs to know the size of the window in which it is running in order to be able to correctly display the installation status. However, when logging into the zone's console using zlogin, this does not pass the number of rows and columns from the window into the zone's login shell as a remote login application such as ssh, telnet, or rlogin would, so the installer fails.

Workaround:

Use a remote login application such as ssh, telnet, or rlogin to login to the zone and run the installation. This correctly sets the terminal rows and columns from the window in which it was launched. Alternatively, use the stty command to set the correct number of rows and columns once you login with zlogin.

Agent Installed in a Non-Global Zone Fails to Register (CR 6511890)

Description:

If an agent is installed in a non-global zone on a system before an agent is installed in the global zone, the agent in the non-global zone will fail to register and a message in the log file will display: ?Host/Group not found?.

Workaround:

Install an agent in the global zone. The non-global zone agent will successfully register after the global zone agent has registered.

Installation of Packages or Patches on Sparse Zone Fails (CR 6508687)

Description:

When attempting to install a package or patch on a sparse zone, the installation will fail if the package or patch needs to add content to /usr directory.

Cause:

This is not a bug in the product, but a feature of zones. Sparse zones have a copy of /usr provided by the global zone which is mounted read-only. Installation of any package (by Sun Connection or manually) which needs to place content in /usr will fail because pkgadd is unable to write to /usr

Workaround:

Add the package to the global zone or use a full zone instead of a sparse zone.

CLI Issues

uce_cli and osc Commands Are Not in the /usr/bin Directory on Solaris

Description:

When the CLI package is installed on Solaris, the convenience links for uce_cli and osc are made in the CLI package's bin directory rather than /usr/bin as it is on Linux.

Workaround:

Add /opt/SUNWuce/cli/bin to the user's PATH shell variable. Both uce_cli and osc will then be available without entering the full path name.

CLI Fails to Run (CR 6456099)

Description:

Sometimes the CLI fails to run.

Workaround:

    Do the following:

  1. Log in to the console as the same user that you used to access the CLI.

    The console should indicate that this user is already logged in.

  2. Confirm this message and then exit console.

    CLI will now be accessible.

SDS Issues

Uploading Packages By Using Upload Web Page Can Hang (CR 6520232)

Description:

When uploading packages using the upload web page, the upload can appear to hang.

Cause:

This only occurs when there is an HTTP proxy between the browser and the SDS and when that proxy has a low connection timeout. The component is uploaded to the SDS correctly, but the browser never receives a response because the proxy times-out the connection.

Workaround:

When uploading packages, disable the use of the proxy in the browser.

RPM Command Is Needed For Provisioning Linux Systems From Solaris SDS

Description:

In order for the Solaris SDS to correctly handle and provision Linux RPM's, it needs to be able to extract information from those RPM's. In order to do this it requires the rpm command, which is not a part of standard Solaris.

Cause:

The rpm command is not a part of the standard Solaris installation.

Workaround:

On the SDS host, install the SFWrpm package from Solaris 10 distribution media or http://www.sun.com/software/solaris/freeware/s10pkgs_download.xml.

Download of Solaris Packages From the Internet to Solaris SDS Can Fail (CR 6512582)

Description:

Using the console, the user can mark packages for download and installation from SunFreeware sites as part of a job. However, when attempting to do this, the job fails and the job log reports something like:Download Phase - started. Failed to download blob 40002677 , Aborting task ! Download Phase - Download Failed. Check server logs for more details. Download in progress. Client should ask again for the file download action.

Workaround:

Once the download has failed, re-run the same job and the package should be installed. If the job contains multiple packages to install, the first time the job attempts to download each package, it will fail, but a re-run should succeed.

Another workaround is to use the Solaris baselines and the baseline pre-caching script to pre-download patches for you. You can download an entire Solaris baseline to your SDS and it allows you to monitor the download for any patch download failures. It also greatly enhances the speed with which future jobs are run on those agent.

For information about how to use baselines, see the following articles:

These articles, along with other useful information, are available in http://www.sun.com/bigadmin/hubs/connection/ on BigAdmin.

Backup Procedure Perform Database Maintenance

Description:

The backup procedure performs database maintenance actions. If you have a large database, the backup might take longer.

Workaround:

You can comment out the maintenance lines. Contact Sun Support for further details.

Backup of Linux SDS Database Fails During Restore on Solaris SDS (CR 6514761)

Description:

It is not possible to migrate from a Linux SDS running Sun Connection 1.0.3 or 1.1 to a Solaris SDS running the 1.1 release. The restore operation on Solaris fails with errors similar to the following:


Stopping UCE Services. This may take a while, please be patient...
 

...Done. Restoring backed up files...
 

tar: ././@LongLink: typeflag 'L' not recognized, converting to regular file
 

tar: ././@LongLink: typeflag 'L' not recognized, converting to regular file
 

tar: ././@LongLink: typeflag 'L' not recognized, converting to regular file
 

...
Workaround:

Sun Connection 1.1.1 includes a workaround for migrating the SDS from the Linux to the Solaris OS.Contact Sun Support for further details.

Console Issues

Editing Host Properties in a Zone Might Cause Database Problems (CR 6545220)

Description:

In Sun Connection 1.1.x, if a system contains zones and you attach a profile to the host properties in a local zone, the system is no longer displayed in the console.

Caution – Caution –

Do not edit host properties in the global zone. If you do, the database will become corrupted and will require manual intervention.

Large Reports Might Cause the Console to Crash

Description:

In Sun Connection 1.0.x, running large reports might crash the console.

uce_console command is not in the /usr/bin Directory on Solaris

Description:

When the console package is installed on Solaris, the convenience link for uce_console is made in the console package's bin directory rather than /usr/bin as it is on Linux.

Workaround:

Add /opt/SUNWuce/console/bin to the user's PATH shell variable. uce_console will then be available without entering the full path name.

Opening the Job Log in the Console Can Take Several Minutes on a Zoned System (CR 6508689)

Description:

When a job is run on multiple agents installed in zones on a system, attempting to open the job log in the console for any one of the jobs can take as long as 10 minutes.

Workaround:

Seekers are still running on the agent which causes the agent to be slow to respond. Once the seekers have finished running, the job log will open in 1-2 seconds. Wait for agents to complete running the seekers.

General Issues

BMI Is Unavailable

Description:

The bare metal installation (BMI) server module is not available.

Workaround:

If you are interested in creating and using a BMI server with Sun Connection, please contact your Sun Sales representative.

SDS Appears to Download Content That Is Already Cached (CR 6456161)

Description:

Sometimes the System Dependency Server appears as if it is downloading content that is already cached and available to the agent.

Workaround:

This is a reporting error in the user interface. Content that is available in the cache is not downloaded again.

VM Cloning Is Not Supported

VM cloning is not yet supported in Sun Connection.

Support Issues

This section describes issues that concern Sun Connection support.

Latest Knowledge Channels and System Requirements

To get a list of available knowledge channels available for Sun Connection and system requirements, go to BigAdmin at http://www.sun.com/bigadmin/hubs/connection/.

Upgrade Issues

This section describes issues that you might encounter while upgrading your Sun Connection software.

Cannot Use the Upgrade Scripts to Upgrade From Version 1.1 to 1.1.1

Description:

The Sun Connection upgrade scripts fail when upgrading from version 1.1 to version 1.1.1.

Workaround:

Use one of the following agent upgrade methods to upgrade from 1.1 to 1.1.1:

  • If you have a Solaris 10 agent, use the special upgrade mode in the console.

    -update_director_mode

  • If you have a Solaris 8 or 9, or Linux agents, contact Sun Support to get the upgrade scripts and procedures to launch a job for upgrade.

Cannot Use the Upgrade Scripts to Upgrade Solaris 10 Agents

Description:

The Sun Connection upgrade scripts fail when upgrading Solaris 10 from version 1.0.x to version 1.1.x.

Note –

The upgrade script does work when upgrading from version 1.1 to 1.1.1.

Workaround:

Contact Sun Support to get the upgrade scripts and procedures to launch a job for upgrade.

Upgrading from Sun Connection 1.0.x to 1.1.x Fails Without Manual Intervention (CR 6531803)

Description:

Upgrading from Sun Connection version 1.0.3 to version 1.1 or 1.1.1 fails and results in a cpio error in the /logs directory.

The failure is caused by a change in the /logs directory between the two releases.

Workaround:

  1. Contact Sun Support to get the upgrade scripts and procedures to launch a job for upgrade.

  2. Rename the server/logs/ directory to server/logs.old

  3. Use the upgrade scripts and procedures to launch the upgrade job.

If you receive the error message, rename the directory and then re-run the job.

Missing Credentials After an Upgrade From Sun Aduva OnStage 439 to Sun Connection

Description:

The credentials you entered to access software directly from a vendor, such as for the Solaris patches, are not saved as part of the backup process.

Workaround:

Add these credentials after the upgrade is complete by using the Sun Connection Authentication window.

Missing Cached Public Components After an Upgrade From Sun Aduva OnStage 439 to Sun Connection

Description:

The cached public components that were downloaded with Sun Aduva OnStage are not saved as part of the backup process.

Workaround:

Download the cached public components after the upgrade is complete.

Attempts to Install Solaris Packages Fail After an Upgrade From Sun Aduva OnStage 439 to Sun Connection (CR 6464386)

Description:

When attempting to install Solaris packages, the package install fails with an error regarding the parameter CLIENT_BASEDIR.

Workaround:

    Do the following:

  1. Copy the Sun Connection agent tar ball to the host.

  2. Remove the ADVagent package.

  3. Unpack the Sun Connection agent tar ball.

  4. Install the Sun Connection agent.

Download the cached public components after the upgrade is complete.

Job Logs not Available After Upgrade From Sun Aduva OnStage 439 to Sun Connection (CR 6454745)

Description:

After the upgrade, job logs from the previous release are no longer available

Workaround:

Backup the logs before the upgrade and restore them after the upgrade.

Before the upgrade, backup the OnStage agent log directory content located at: /opt/local/aduva/director_agent/log .

After the upgrade, restore the saved data into the /opt/local/uce/agent/log directory, which is Sun Connection's agent log directory.

Configuration Issues

This section describes issues that you might encounter while configuring your Sun Connection environment.

Enabling Log Rotation in the Sun Connection Proxy (CR 6455904)

To enable log rotation in the Sun Connection proxy, manually install a copy of the Log_rotation configuration file on the proxy.

The file is on the server at /etc/logrotate.d/director_server.

Cookies are Expired When Downloading Red Hat Components (CR 6456196)

Description:

When you attempt to download Red Hat components from the Red Hat Network, the download fails because the cookies have expired.

You might see this problem when you use other web clients, such as Konqueror.

Cause:

This problem occurs when the system clock on your SDS is not set to the current time, but is set forward to a time in the future. Cookies created on your system might be expired.

Workaround:

Ensure that the system clock on the SDS machine is set to the current time. The system clock must be set to the current time because it is used for cookie validation.

Sun Connection Application Issues

This section describes issues that you might encounter when using Sun Connection applications.

Solaris Channels Cannot Handle Patches Uploaded In A *.tar.gz Format (CR 6571151)

Description:

The browser-based upload utility, upload.html, allows you to upload patches in various packed formats (*.zip, *.tar, *.tar.gz). However, the agent can only handle patches in a *.zip or **.jar format. The agent cannot handle a patch in the *.tar.gz format.

Results:

An error message similar to the following is displayed:

Execution of PATCH Job FAILED. Details: PATCH_INSTALL_JOB_124630-06-0
Patch intaller failed. Check if should continue.
Task will continue.
Commit Phase - Finished With error: 117515008 ( Cannot run job. ).
} Task Finished with a failure ! ( JobID = 52 )
Workaround

If the patch cluster is in a *.tar.gz format, unpack the archive and then use a *.zip or *.jar format to repack the patches.

Time Out Message When Running a Large Job In Single-User Mode

Description:

If you are single-user mode and running a large job on a Solaris 10 system, you might see the following message on the serial console:


svc.startd[7]: svc:/milestone/single-user:default: Method or service exit timed out.
  Killing contract 27.
[ milestone/single-user:default timed out, fault threshold reached
  (see 'svcs -x' for details) ]
Requesting System Maintenance Mode
(See /lib/svc/share/README for more information.)
Console login service(s) cannot run
Root password for system maintenance (control-d to bypass):
Workaround:

This is normal. The patch job is still running, and no action is required.

To prevent the message from displaying, increase the default single-user milestone time out limit. The timeout seconds parameter is located in the /var/svc/manifest/milestone/single-user.xml file.

Console Sometimes Hangs (CR 6456138)

Description:

Sometimes the console hangs if open for a long time.

Workaround:

Close the console when not is use or close the console occasionally when you use it for a long time.

The system on which you run the console should have at least 1 Gbyte of memory.

Console Can Hang When Preferences Changed (CR 6464925)

Description:

If the console preferences are changed, and then an attempt is made to change them for a second time, the console can hang.

Workaround:

Kill the console and restart it.

Job Can Hang When "Remote Hosts" is Used to Upload RPM (CR 6466850)

Description:

If the "Remote Hosts" option is used to upload a file, and the remote host goes offline during the upload, the job remains stuck in the console, even after the remote host restarts.

Workaround:

Kill the console and restart it.

Copying Non-Certified Objects Between Channels (CR 6456000)

If you want to copy Non-Certified Objects (NCOs) between channels, use the scriptable API.

Managed Host Issues

This section describes issues that impact Sun Connection managed hosts.

Reboot Notification Flags are Not Set When Host is Marked as Pending Reboot (CR 6464388)

Description:

    When a user chooses not to perform an immediate reboot of a host, the Pending Reboot is indicated by the Sun Connection console twice:

  1. In the Host progress dialog.

    There is a line on the Host's TODO list saying Remember to reboot

  2. In the Inventory window.

    The inventory window displays the host list on the left and the components tree on the right.

As part of the components tree there is a specific root category named Notifications. Within this category there are two components: Reboot and Reboot Reconfigure. These two components are normally marked as (installed) when the selected host has 'Pending Reboot'. However, this indication is not working. The (installed) indication is not shown.

Workaround:

This does not affect the operation of the application.

Documentation Issues

The following are the major changes between Sun Update Connection – Enterprise and Sun Connection:

Location of Sun Connection Documentation

Sun Connection documentation is available on BigAdmin at http://www.sun.com/bigadmin/hubs/connection/. Articles regarding how to install, configure, and use Sun Connection Satellite, formerly known as Sun Update Connection – Enterprise, are limited. As articles are written, they will be published to the Sun Connection site on BigAdmin. If you are unable to find an article for the topic that you want, see the Sun Update Connection – Enterprise documentation, which is available at http://docs.sun.com/app/docs/coll/1561.1.

The /usr/local/uce/cli/bin/pkg_loader.sh Script No Longer Exists (CR 6493707)

Description:

The pkgloader.sh script is no longer available. In addition, the CLI's directory location has changed.

Workaround:

The equivalent functionality for the pkg_loader.sh script is available through the CLI (upload_files). This functionality is not available on Windows systems.

The CLI directory location has changed to the following:

  • Solaris – /opt/SUNWuce/cli/bin/uce_cli

  • Linux – /usr/bin/uce_cli