Chapter 9 Known Limitations and Workarounds

Table of Contents

Find information on known limitations and workaround procedures for Oracle VM.

The titles of known issues contain platform designations as follows:

  • If an issue applies to Oracle VM Server for x86 only, the title includes x86 Only.

  • If an issue applies to Oracle VM Server for SPARC only, the title includes SPARC Only.

  • If an issue applies to both Oracle VM Server and Oracle VM Server for SPARC, the title does not include any platform designation.

9.1 

9.1.1 Integration Between Oracle VM Server for SPARC and Oracle VM Manager

In addition to known issues that might apply for a given release of Oracle VM, limitations exist for managing Oracle VM Server for SPARC with Oracle VM Manager.

There are currently several features in Oracle VM Server for SPARC that cannot be represented in Oracle VM Manager. These features include the following:

  • Physical I/O domains such as SR-IOV, direct I/O, and root domains.

  • Certain network options such as Private VLAN (PVLAN), "vNICs on VNets" or multiple MAC addresses per virtual network device, and Quality of Service (QoS) configurations.

  • Crypto accelerators assigned to guest domains. This is a feature on SPARC servers prior to the T4 series that required some administrative effort. With the T4 series and later, crypto accelerators are always available without any administrative effort required.

  • Whole-core and max-cores constraints, although these can be configured by editing the configuration file named /etc/ovs-agent/agent.ini that is used by the Oracle VM Agent for SPARC on each Oracle VM Server.

  • Multiple service domains, also known as 'shadow service domains', which are typically used to provide redundancy. Initial support for these has been implemented in the Oracle VM Agent for SPARC.

There are a number of limitations that exist in Oracle VM Manager for server pools that contain Oracle VM Server for SPARC servers. These limitations include the following:

  • ISCSI or Fibre Channel storage cannot be used to store a repository or a server pool cluster file system. However, these disks can be used for direct virtual machine assignment. NFS storage should be used if you need to set up a server pool cluster file system. Either NFS, ZFS or a local disk may be used to configure a repository.

  • Huge page support on SPARC systems is not controlled by Oracle VM. All the page sizes available on the underlying hardware are always available to the guest virtual machine.

  • Virtual machine suspend and resume functionality is not available for virtual machines running in a SPARC-based server pool.

  • Support for a secondary service domain, running alongside the primary service domain, has been added to the Oracle VM Agent. This feature allows you to restart the primary service domain without causing any interruption to network or disk I/O for any of your running virtual machines.

9.2 Known Issues and Workarounds

9.2.1 After Upgrading To Release 3.4.6 Unable To Log Into the Oracle VM Manager Web Interface Using Custom Certificates

During an upgrade to Oracle VM Manager Release 3.4.6 files in /u01/app/oracle/ovm-manager-3/domains/ovm_domain/security are cleaned and only files created and managed by Oracle VM Manager are preserved and restored. Files created by users are not preserved, except for relevant keystore files.

Workaround: For more information, see My Oracle Support (MOS) article 2629033.1 at http://metalink.oracle.com.

Bug 30365865

9.2.2 Oracle VM Server Virtual Machine Fails To Reboot After Saving vmcore

The kdump reboot may fail on Oracle VM Manager Release 3.4 if the gap between kdump boot and kdump reboot is less than 10s.

Workaround: There is no workaround for this issue.

Bug 30995725

9.2.3 Server Operation Succeeded But Job Failed Due to a Late Lock Exception

Occasionally, during Oracle VM Server operations a late lock exception error may appear.

Workaround: To resolve this problem, refresh the storage repository and the server. For more information, see MOS article 1672620.1 at http://metalink.oracle.com.

Bug 31014915

9.2.4 Upgrading Oracle VM Agent for SPARC Release 3.3.x/3.4.1/3.4.2 to Release 3.4.6 Fails

Upgrading Oracle VM Agent for SPARC Release 3.3.x, 3.4.1, or 3.4.2 to Release 3.4.6 results in errors and eventually fails to upgrade the server.

Workaround: The workaround is to first upgrade Oracle VM Agent for SPARC Release 3.3.x, 3.4.1 or 3.4.2 to Release 3.4.5 on Oracle Solaris 11.3. From Release 3.4.5, after a reboot you can then upgrade to Release 3.4.6 with Oracle Solaris 11.4. For more information, refer to the Oracle VM Installation and Upgrade Guide.

Bug 28902444

9.2.5 Editing a Bond Port Fails on Oracle VM Server for SPARC Running Oracle Solaris 11.4

Editing a bond port fails on Oracle VM Server for SPARC running Oracle Solaris 11.4 with an command error message similar to the following: 

/usr/sbin/dladm: -m is no longer supported.

This failure occurs because of a change in Oracle Solaris 11.4.

Workaround: The workaround is to remove the targeted port from the network and to delete the VLAN interfaces using the port so that this bond port can be deleted and recreated with the appropriate mode.

Note

A running virtual machine that is using the network or port must either be stopped or migrated to another server before the workaround operation can be performed.

Bug 28821715

9.2.6 Cannot Install or Upgrade Oracle VM Manager Release 3.4.5 or 3.4.6 on Oracle Linux 7 Update 6 Guest Virtual Machines

Upgrading Oracle VM Manager Release 3.4.5 to 3.4.6 fails on an Oracle Linux guest virtual machine when the guest OS is first upgraded to Oracle Linux 7 Update 6. Similarly, installing Oracle VM Release 3.4.6 on guest virtual machines running Oracle Linux 7 Update 6 is failing.

Workaround: Upgrade the guest OS to Oracle Linux 7 Update 6 after Oracle VM Manager Release 3.4.6 is running on the virtual machine.

Bugs 28907767 and 28898808

9.2.7 Attached Disks Not Visible at Startup on Oracle Solaris 11.4 Virtual Machine

After creating a Oracle Solaris 11.4 virtual machine on Oracle VM Release 3.4.6, attached disks are not visible at startup when running the format command.

Workaround: The workaround is to run the format command from the /dev/rdsk/ directory to display all attached disks.

Bugs 28797606

9.3 Known Issues and Workarounds

9.3.1 Cannot Re-install Oracle VM Server Across Multiple Disks or LUNs using a Kickstart Configuration

If you try to re-install Oracle VM Server using a kickstart configuration across multiple LUNS, with one LUN dedicated to swap, the installation fails with a device error message, similar to the following: 

DeviceError: ('cannot replace active format', 'sdh')

This occurs because during the re-installation, the anaconda installation wizard fails to reformat the LUN dedicated to swap as it is already in use from the previous installation, resulting in the error.

Workaround: There is no workaround for this issue. Oracle VM Release 3.4 supports installations of Oracle VM Server onto a single disk or LUN only. As a result, kickstart installations across multiple disks in a non-software RAID configuration is not supported.

Bug 27021529

9.3.2 Oracle VM Server Installation Hangs with RAID10 Boot Partition

If you install Oracle VM Server on a machine with a RAID10 disk configuration, the installation process hangs at the end, because the command mdadm --wait-clean --scan cannot be completed. This issue is caused by the fact that boot partitions are not supported on RAID10, but only on RAID1.

Workaround: Configure the software RAID on the server such that the /boot partition is written to a RAID1 device. Restart the Oracle VM Server installation on the supported configuration.

Bug 27525377

9.3.3 Cannot Retrieve Information for File System Mounted on Invalid Mount Point

If you are utilizing Oracle VM Server for SPARC in your Oracle VM environment and you are mounting a file system, you must ensure that no white spaces or invalid characters are specified in the mount point directory path. This also applies when mounting devices using the automounter mechanism. Mounting file systems on invalid mount points can result in file system information retrieval failures, similar to: 

AttributeError: 'NoneType' object has no attribute 'split'

Workaround: If possible, rename the device or alternatively, unmount the invalid mount point and mount the device to a mount point that contains no white spaces or invalid characters. Refer to the Oracle Solaris 11.3 documentation for information on managing file systems: https://docs.oracle.com/cd/E53394_01/html/E54785/.

Bug 27479482

9.3.4 Cannot Configure the Oracle Solaris Support Repository using the Oracle VM Manager Web Interface

If you are updating a set of local Oracle VM Servers running on SPARC hardware, you may want to utilize the Oracle Solaris Support Repository with it's associated SSL key and certification. Currently, the Oracle VM Manager Web Interface does not provide any option to insert the required key and certification when creating a new server update repository.

Workaround: Instructions on installing a certificate and key can be found in the book titled Install a Certificate and Key for the Oracle Solaris Support Repository in the Oracle Solaris 11.3 Information Library at: https://docs.oracle.com/cd/E53394_01/html/E54747/installkeycert.html

Bug 27572483

9.3.5 Backup Failure for the MySQL Database Repository

If you specify a new MySQL database backup directory, you may encounter a failure during the MySQL database backup if the new directory does not exist, or does not have the correct permissions assigned. By default, the MySQL database backup files are stored within /u01/app/oracle/mysql/dbbackup on the Oracle VM Manager host, but it is possible to set a different directory if required.

Workaround: When changing the default MySQL database backup directory, you must ensure that the new directory exists and has the correct permissions assigned on the Oracle VM Manager host as follows:

  1. Create the new directory that you want to use for the MySQL database backup files, for example: 

    # mkdir /u01/app/oracle/mysql/new-backup-dir

  2. Ensure the right permissions are set for the new directory: 

    # chmod 777 /u01/app/oracle/mysql/new-backup-dir

  3. Update /etc/sysconfig/ovmm and specify the new directory for the DBBACKUP parameter: 

    DBBACKUP=/u01/app/oracle/mysql/new-backup-dir

  4. Save and close /etc/sysconfig/ovmm for the new directory to take effect.

For more information on backing up the MySQL Database repository, see Backing up and Restoring Oracle VM Components in the Oracle VM Administrator's Guide.

Bug 27582009

9.3.6 Unable to Login into the Oracle VM Web Services API

Connections to the Oracle VM Web Services API fail if you are using a Java runtime version lower than 1.7.0_131_b31. As of Oracle VM Release 3.4.5, the SSL TLS version 1 (TLSv1) protocol is disabled within Oracle VM Manager for security reasons and connections using the TLSv1.2 protocol are only accepted. The TLSv1.2 protocol is the default in Java runtime version 1.7.0_131-b31, and as a result, connections to the Oracle VM Web Services API fail if a lower Java runtime version is used.

Workaround: Make sure you are using Java runtime version 1.7.0_131-b31 or higher. The Java runtime version that ships with Oracle VM Manager Release 3.4.5 is sufficient. Alternatively, for lower versions of Java runtime, update the Java client and make sure that the SSLContext ctx attribute is set as follows: 

SSLContext ctx = SSLContext.getInstance("TLSv1.2");

For more information about authentication against the REST API in the Java Client, see the Java topic in the Oracle VM Web Services API Developer's Guide.

Bug 27611305

9.3.7 Incorrect PVM Migration Error Message Displayed when Secure VM Migration Enabled

As of Oracle VM Release 3.4.5, you can disable PVM guests from running on your Oracle VM Server instance. As a result, migrating PVM guests to a destination Oracle VM Server instance with PVM guests disabled is not possible and the following error message is displayed: 

stderr: Error: PV guests disabled by xend.

However, when migrating PVM guests with secure vm migration enabled to a destination Oracle VM Server instance with PVM guests disabled, the following incorrect message is displayed: 

stderr: Error: [Errno 9] Bad file descriptor

Workaround: The incorrect migration error message should be ignored. The PVM migration is still prevented regardless of whether secure vm migration is enabled or not for the server pool where the vm resides.

For information on how to disable PVM guests on your Oracle VM Server instance, see Disabling Paravirtualized Guests on Oracle VM Server in the Oracle VM Administrator's Guide. For information on secure VM migrations when creating a server pool, see the Create Server Pool topic in the Oracle VM Manager User's Guide.

Bug 27679547

9.3.8 Default VM Start Policy Is Different in Oracle VM Manager Command Line Interface and Oracle VM Manager Web Interface

In Oracle VM Release 3.4.5, the start policies for virtual machines have changed. At the server pool level, the Oracle VM Manager Command Line Interface and Oracle VM Manager Web Interface apply a different default start policy: a server pool created through the web UI uses the "Best Server" policy, while a server pool created through the CLI uses the "Current Server" policy.

Workaround: There is no functional issue. The difference in behavior between both interfaces will not be addressed, because changing the default policy may break existing customers' CLI automation. Set the virtual machine start policy within your server pools as required.

Bug 27805945

9.3.9 Virtual Machine Fails to Boot with kexec

If you attempt to boot a virtual machine into a different kernel using kexec, the sequence sometimes fails due to certain devices not loading. Issues have been reported with missing networking or logical volumes, on a variety of kernel combinations. No root cause has been identified.

Workaround: The issue is intermittent, so you can retry. There is currently no workaround.

Bug 27338786

9.4 Known Issues and Workarounds

9.4.1 Missing Physical Disk Warning for Oracle VM Server Instance After Upgrading Oracle VM Manager

If you are using Oracle VM Manager Release 3.4.2 or later to manage previous versions of Oracle VM Server that have not yet been upgraded, warnings about missing physical disks in the internal ZFS storage appliance may be displayed in the Oracle VM Manager Storage tab and the Physical Disks perspective of the Servers and VMs tab. For example: 

OVMEVT_007005D_000 Discover storage elements on server [ovcacn10r1] did not return
physical disk [SUN (2)] for storage array [OVCA_ZFSSA_Rack1].

Workaround: Upgrade all instances of Oracle VM Server to the same release version of Oracle VM Manager.

Bug 26320569

9.4.2 Suspend Error During Storage Live Migration

When migrating running virtual machines that use local storage, the virtual disks are migrated first from the source server's local repository to the destination server's local repository. In certain cases, during the storage live migration, a suspend error may occur for the source Oracle VM Server guest virtual machine when there is additional disk activity. For example, additional disk activity may involve a process that is running on the guest virtual machine that is writing to the virtual disk on the source server's local repository while it is being migrated to the destination server's local repository.

The suspend operation requires more time to complete due to the additional disk activity, resulting in a suspend error that causes the storage live migration to fail. A suspend error similar to the following is displayed in the /var/log/xen/xend.log on the source Oracle VM Server instance: 

[timestamp 2792] DEBUG (XendDomainInfo:3509) waitForSuspend: domain 1 state:
[timestamp 2792] INFO (XendCheckpoint:430) [9836] xc: error:
Suspend request failed: Internal error
[timestamp 2792] DEBUG (XendDomainInfo:1943) XendDomainInfo.handleShutdownWatch
[timestamp 2792] INFO (XendCheckpoint:430) [9836] xc: error: Domain appears not
to have suspended: Internal error

A timeout error is also displayed in the kernel log on the source Oracle VM Server guest virtual machine: 

timestamp server kernel: [ 1601.227167] Freezing of tasks failed
after xx seconds (0 tasks refusing to freeze, wq_busy=1):

Workaround: Update the /sys/power/pm_freeze_timeout parameter, which is specified in milliseconds on the source Oracle VM Server guest virtual machine to ensure adequate time is available when additional disk activity occurs during storage live migration:

  1. Log in to the guest virtual machine.

  2. It is strongly recommended to update the pm_freeze_timeout parameter to the maximum downtime that you can afford, up to a maximum of 5 minutes, or 300000 milliseconds. For example, to update the /sys/power/pm_freeze_timeout parameter to 5 minutes, run the following command on the guest virtual machine: 

    # echo "300000" > /sys/power/pm_freeze_timeout

    Note

    The value you set for the pm_freeze_timeout parameter impacts the amount of time the source Oracle VM Server guest virtual machine remains offline during storage live migration.

  3. Perform the live migration with storage again and no suspend errors should occur.

If a suspend error still occurs for the source Oracle VM Server guest virtual machine after the pm_freeze_timeout parameter is updated, you may need to avoid storage live migration until the additional disk activity completes, or reconsider the maximum time you can afford for the guest virtual machine to remain offline during storage live migration.

Bug 26289880

9.4.3 Cannot Change Memory Configuration of Running PVHVM Guests Based on CentOS 6

For PVHVM guests based on CentOS 6, the available memory is set to the maximum memory in the virtual machine configuration when the virtual machine starts. Any operations to modify the memory while the guest is running complete successfully but do not take effect.

Workaround: Stop the virtual machine and then change the memory configuration.

Bug 26542210

9.4.4 PVHVM Guests Based on CentOS 5 Do Not Start Successfully After Conversion from PVM

Note

As of Release 3.4.6, support for PVM guests is removed. For information information, see Section 8.1, “Removed Features in 3.4.6”.

If you change the domain type of a virtual machine based on CentOS 5 from PVM to PVHVM, and then attempt to start the virtual machine, it does not boot successfully. The following error displays in the virtual machine console: 

Error 13: Invalid or unsupported executable format

Workaround: There is no workaround for this issue.

Bug 26537271

9.4.5 PVM Guests Based on CentOS 5 Do Not Start Successfully After Conversion from PVHVM

Note

As of Release 3.4.6, support for PVM guests is removed. For information information, see Section 8.1, “Removed Features in 3.4.6”.

If you change the domain type of a virtual machine based on CentOS 5 from PVHVM to PVM, and then attempt to start the virtual machine, it does not boot successfully. The following error message is displayed: 

ELF image has no shstrtab

Workaround: There is no workaround for this issue.

Bug 26537152

9.4.6 Guests Based on CentOS 5.11 Start with the Maximum Memory Configuration

Virtual machines based on CentOS 5.11 start running with a total memory that is equal to the maximum memory instead of the available memory, as defined in the virtual machine configuration.

Workaround: Change the memory configuration in the running virtual machine.

Bug 26536834

9.4.7 Guests Based on CentOS 5 Display Incorrect Number of VNICs

If you create a virtual machine based on CentOS 5 and then add a VNIC to the virtual machine, that VNIC is duplicated when you reboot the virtual machine. For example, if you add one VNIC, duplicate settings are displayed for the eth0 and eth1 interfaces.

The duplicate VNIC is caused by an issue with the CentOS 5 kernel that loads two drivers for the same network interface.

Workaround: There is no workaround for this issue.

Bug 26535868

9.4.8 Live Migrating Oracle Solaris Guests to Oracle VM Server Release 3.4.4 or Later Results in Guest Reboot

Live migrating Oracle Solaris HVM or PVHVM 64-bit guests from previous versions of Oracle VM Server to a destination Oracle VM Server at Release 3.4.4 or later results in an Oracle Solaris guest reboot. The following message is quickly displayed on the Oracle Solaris guest console, followed by a reboot: 

gnttab_maps: Invalid pfn

Workaround: The following migration options are available:

Live migrate Oracle Solaris HVM or PVHVM 64-bit guests to Oracle VM Server Release 3.4.4 or later

Note

The following points should be reviewed before proceeding with this solution:

  • To ensure minimal disruption, it is recommended to complete the instructions described in this solution immediately after upgrading to Oracle VM Server Release 3.4.4 or later and before the Oracle VM Server is returned to service.

  • All Oracle VM Server instances at Release 3.4.4 or later in the server pool must be updated based on the instructions described in this solution.

  • The configuration setting update described in this solution will be carried forward during future Oracle VM Server upgrades.

  1. Using the Oracle VM Manager Web Interface, edit the destination Oracle VM Server and place it into maintenance mode. For more information, see Edit Server.

    Note

    If you have just upgraded to Oracle VM Server Release 3.4.4 or later, the Oracle VM Server may already be in maintenance mode.

  2. On the Oracle VM Server instance, open /etc/default/grub for editing.

  3. Add gnttab_max_frames=64 to the GRUB_CMDLINE_XEN line.

  4. Regenerate the GRUB2 configuration.

    • BIOS-based systems: 

      grub2-mkconfig -o /boot/grub2/grub.cfg

    • UEFI-based systems: 

      grub2-mkconfig -o /boot/efi/EFI/redhat/grub.cfg

  5. Reboot the Oracle VM Server instance for the update to take effect.

  6. Edit the Oracle VM Server on the Oracle VM Manager Web Interface and take it out of maintenance mode by unchecking the Maintenance Mode check box.

  7. Proceed to live migrate the Oracle Solaris guest to the destination Oracle VM Server instance.

Migrate Oracle Solaris HVM or PVHVM 64-bit guests to Oracle VM Server Release 3.4.4 or later

Note

Although live migration of Oracle Solaris guests to Oracle VM Server Release 3.4.4 or later is not possible with this solution, you avoid updating the configuration setting on the Oracle VM Server instance and the consequent required server reboot. In addition, there will be improved guest performance and the custom parameter gnttab_max_frames=64 will not need to be maintained on any future installations of Oracle VM Server.

  1. Before migrating the Oracle Solaris guest to the destination Oracle VM Server, you must stop the virtual machine guest on the source server.

  2. Migrate the Oracle Solaris guest to the destination Oracle VM Server.

  3. Once migration completes, restart the Oracle Solaris guest on the destination Oracle VM Server.

For more information, see Migrate or Move Virtual Machines.

Bug 26594820

9.5 Known Issues and Workarounds

9.5.1 Cannot Remove Software RAID Devices from Oracle VM Manager

You cannot use Oracle VM Manager to remove physical disks that are included in a software RAID. If you attempt to remove software RAID devices from Oracle VM Manager, the job fails with the following error message: 

OVMRU_007016E: Operation is not supported for local storage.

Workaround: Manually remove the software RAID devices from Oracle VM Server. For more information, see Removing Software RAID Devices in the Oracle VM Administrator's Guide.

Bug 25554732

9.5.2 VNICs Deleted When Editing Virtual Machine

Modifying a virtual machine configuration deletes the VNICs associated with that virtual machine if edit jobs are in progress. For example, if you modify a virtual machine and then perform a subsequent modification before the first job completes, the VNICs associated with that virtual machine are deleted.

Workaround: When editing a virtual machine, you must wait until all previous edit jobs complete before making any further modifications.

Bug 24961613

9.5.3 Enable High Availability Option is Disabled Due to Restart Option

If you edit a virtual machine configuration to enable high availability, and the Restart or Restart After Dump option is set to either Stop or Stop After Dump, the Enable High Availability option is disabled on the Oracle VM Manager Web Interface.

Workaround: To enable the Enable High Availability option, you must first edit the virtual machine configuration and set the Restart Action on Crash option to Restart or Restart After Dump and then save the configuration before you attempt to enable high availability.

Bug 24961613

9.5.4 Cannot Upgrade Oracle VM Manager Due to Presence of Temporary Files

When upgrading Oracle VM Manager Release 3.3.3 to Release 3.4.x, the upgrade procedure may fail due to the presence of temporary files.

Workaround: Delete any temporary files that cause the upgrade failure. For example, run the following command as the root user on the Oracle VM Manager host: 

# rm -Rf /tmp/workdir_sel
# rm -Rf /tmp/ovm*

Once these files are removed, restart the upgrade procedure again.

Bug 25394965

9.5.5 Job List Excel Spreadsheet Name Contains Unexpected Characters

In the Oracle VM Manager Web Interface, when you export the job list as a Microsoft Office Excel spreadsheet, the browser Save/Open File dialog may display the Job List Excel spreadsheet name with unexpected characters, for example, =_UTF-8_Q_JobList=2Exls_=. This issue occurs if you are using Microsoft Internet Explorer 11 as your browser, and you are using a language setting other than English.

Workaround: Although the Excel spreadsheet can be successfully downloaded and opened, it is recommended to remove the unexpected characters and rename the spreadsheet to JobList.xls.

Bug 25574313

9.5.6 Oracle Linux 6 HVM Guest Virtual Machine with Hard Disk of 2TiB in Size Hangs on Boot

You cannot start hardware virtualized (HVM) guests running Oracle Linux 6 if the guest boots from hard disks of 2TiB in size.

Workaround:

  1. Host the contents of the Oracle VM Server ISO image file over the network so that you can access the required GRUB2 package files.

  2. Log in to the Oracle VM Manager Web Interface.

  3. Clone the Oracle Linux 6 virtual machine and ensure that it is not attached to a hard disk of 2TiB in size.

  4. Attach the 2TiB hard disk to the cloned virtual machine.

  5. Install the grub2 packages on the 2TiB hard disk that you attached to the cloned virtual machine, for example: 

    # cd /tmp
# wget --no-proxy /path/grub2-version.rpm
# wget --no-proxy /path/grub2-tools-version.rpm
# mkdir tmp
# cd tmp
# rpm2cpio ../grub2version.rpm | cpio -iumd
# rpm2cpio ../grub2-tools-version.rpm | cpio -iumd
# /usr/sbin/grub2-install -d ./usr/lib/grub/i386-pc/ --force /dev/sdb
  Installing for i386-pc platform.
  ./usr/sbin/grub2-install: warning: this GPT partition label contains no BIOS Boot 
  Partition; embedding won't be possible.
  ./usr/sbin/grub2-install: warning: Embedding is not possible. GRUB can only be 
  installed in this setup by using blocklists. However, blocklists are UNRELIABLE and 
  their use is discouraged.
  Installation finished. No error reported.

    Note

    • Substitute path with the path to the contents of the Oracle VM Server ISO image file on your file system.

    • Substitute version with the grub2 package version contained in your Oracle VM Server instance.

    • Substitute /dev/sdb with the device name for the 2TiB hard disk that you attached to the cloned virtual machine.

  6. Create the /boot/grub2/grub.cfg configuration as follows: 

    set default="0"
set timeout=5
menuentry 'Linux' {
    insmod part_gpt
    insmod ext2
    set root='hd0,gpt1'
    linux /vmlinuz-version ro root=/dev/mapper/VolGroup-lv_root
    initrd /initramfs-version.img
}

    Note

    Update the vmlinuz and initramfs version accordingly based on the Oracle Linux 6 version installed.

  7. Re-attach the 2TiB hard disk to the virtual machine that you could not boot.

Bug 25655807

9.5.7 Kernel Panic when Assigning More Than 32 virtual CPUs to Oracle Linux 32-bit PVM Guests

Note

As of Release 3.4.6, support for PVM guests is removed. For information information, see Section 8.1, “Removed Features in 3.4.6”.

You cannot assign more than the 32 virtual CPUs to Oracle Linux 32-bit Paravirtualized (PVM) guests running Oracle Unbreakable Enterprise Kernel (UEK) Release 2. Oracle Linux 32-bit PVM guests running UEK Release 2 support a maximum of 32 virtual CPUs. If you assign more than the maximum virtual CPUs allowed, kernel panic may occur.

Workaround: Set the maximum virtual CPUs to 32 or less. Use the uname command to determine the UEK version you are running on the guest virtual machine and from the result, you can determine the maximum number of virtual CPUs supported, for example: 

# uname -r
2.6.39-400.278.3.el6uek.i686
# cat config-2.6.39-400.278.3.el6uek.i686 | grep CONFIG_NR_CPUS
CONFIG_NR_CPUS=32

Bug 25745578

9.5.8 Live Migration of Guest Virtual Machine Fails Due to Linux Kernel Version

Running Linux UEK x86_64 2.6.39-400.109.1.el6 as the guest kernel may cause issues during live migration. For example, live migration may fail with an unresponsive guest virtual machine on the VNC Console.

Workaround: Upgrade the guest kernel to the latest version.

Bug 25644455

9.5.9 PVHVM Guest Boot Delayed by Xen Balloon Driver

Booting PVHVM guests takes longer than expected and the following message is written to stack trace: 

INFO: task <task_name>:234 blocked for more than 120 seconds
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

This issue occurs when the virtual machine configuration includes a maximum memory value that is much larger than the available memory value. This configuration causes the Xen balloon driver to shrink the guest memory to the target size. As a result, the boot time for the virtual machine takes longer than expected.

Workaround: The greater the difference between the values for maximum memory and available memory, the longer it takes to boot PVHVM guests. If you encounter this issue, you should either decrease the maximum memory or increase the available memory in the virtual machine configuration.

Bug 25709221

9.5.10 Oracle VM Server Cannot Boot from Software RAID Devices with Disks that are 2TB or Larger

You should not attempt to install Oracle VM Server on a software RAID if a disk that is 2 TB or larger is a member of the array. Due to a limitation with the Master Boot Record (MBR), Oracle VM Server cannot boot from disks that are 2 TB or larger in a software RAID configuration.

Workaround: There is no workaround for this issue.

Bug 25608136

9.5.11 Oracle VM Server Does Not Boot After Installation to LUN Targets Larger Than 2TB

In some cases, Oracle VM Server does not have a valid boot entry after installation to Fibre Channel or Fiber Channel over Ethernet (FCoE) LUNs. As a result, Oracle VM Server cannot boot successfully. This issue affects both BIOS and UEFI-based environments and is the result of outdated firmware for the host bus adapter (HBA) or other host controller on the target server.

As of Release 3.4.3, support for installation of Oracle VM Server to systems where Fibre Channel or FCoE LUNs are greater than 2TB in size applies to the following tested versions only:

Server

Host Adapter or Controller

Part Number

Firmware version

BIOS or UEFI

Oracle Server X4-4

Sun StorageTek Dual 8 Gb Fibre Channel PCIe HBA

SG-XPCIE2FC-QF8-Z

8.05.00

UEFI

Oracle Server X5-2

Sun Storage 16 Gb Fibre Channel PCIe Universal HBA, QLogic

7101674

8.05.00

UEFI

Oracle Server X5-4

Sun Storage 16 Gb Fibre Channel PCIe Universal HBA, Emulex

7101684

11.1.160.6

BIOS and UEFI

Oracle Server X5-8

Sun Storage 16 Gb Fibre Channel PCIe Universal HBA, QLogic

7101674

8.05.00

UEFI

Note

  • Legacy BIOS SAN boot is not supported for Fibre Channel or FCoE LUNs that are greater than 2TB in size for QLogic HBA or other host controllers on the target server.

  • UEFI SAN boot for Qlogic HBA or other host controllers are supported, but the firmware for the HBA or other host controllers on the target server may need to be updated.

  • Both Legacy BIOS and UEFI SAN boot for HBA or other host controllers, such as Emulex are supported, but the firmware for the HBA or other host controllers on the target server may need to be updated.

Workaround: Update the firmware for the host bus adapter (HBA) or other host controller on the target server.

Bug 25701247

9.5.12 Oracle VM Server Boot Delay and I/O Error Messages with Oracle ZFS Storage Array

A known issue with the Oracle ZFS SCSI storage driver causes significant delays during Oracle VM Server boot and writes a large number of I/O error messages to the kernel log. This issue occurs because the storage driver loads before the SCSI device handler modules (scsi_dh) are available. As a result, the system issues I/O operations to SCSI multipath devices before the devices can handle them.

Workaround: Add the SCSI device handler modules (scsi_dh) to the kernel command line parameters as follows:

  1. Open /etc/default/grub for editing.

  2. Add rdloaddriver=scsi_dh_alua to the GRUB_CMDLINE_LINUX line.

  3. Regenerate the GRUB2 configuration.

    • BIOS-based systems: 

      grub2-mkconfig -o /boot/grub2/grub.cfg

    • UEFI-based systems: 

      grub2-mkconfig -o /boot/efi/EFI/redhat/grub.cfg

Bug 25686610

9.5.13 FIPS Integrity Error Message for SSH Connections from Oracle VM Server

If you are logged in to Oracle VM Server and you establish an SSH connection to another system, the following message is displayed: 

FIPS integrity verification test failed.

This message indicates that Oracle VM Server cannot enable FIPS mode for outbound SSH connections. However, Oracle VM Server can still establish an SSH connection to the remote system but that connection is not validated by FIPS cryptographic algorithms.

Workaround: There is no workaround for this issue.

Bug 25831585

9.5.14 Incorrect MAC Address Displayed for Bond Port

In some cases, Oracle VM Manager displays an incorrect MAC address for bond ports instead of the MAC address assigned for the bond port on Oracle VM Server for SPARC.

Workaround: Ensure the bond port is up by running the dladm show-aggr -x command on the primary domain of the SPARC server, and then rediscover the instance of Oracle VM Server for SPARC.

Bug 25768560

9.6 Known Issues and Workarounds

9.6.1 Kernel Panic when Booting Guest Virtual Machine with 100 or more Virtual Disks

Adding 100 or more virtual disks to a guest virtual machine and then booting the virtual machine with a UEK2, UEK3, or UEK4 kernel results in a kernel panic, as the virtual machine does not recognize all the virtual disks and loses network connectivity.

Workaround: For more information on this issue and the workaround, see Oracle VM: VM Guest With 100+ Virtual Disks Does Not Recognize Them All And Loses Connectivity Document ID 2038050.1, on My Oracle Support at: https://support.oracle.com/oip/faces/secure/km/DocumentDisplay.jspx?id=2038050.1.

Bug 24428913

9.6.2 Different Default Values for the serverAffinityType when Creating Anti-Affinity Groups

When you create an anti-affinity group, the default value for the affinity group type that is returned for the getServerAffinityType method is different, depending on whether you create the anti-affinity group in the Oracle VM Manager Web Interface or the Oracle VM Web Services API. If you create the anti-affinity group in the Oracle VM Manager Web Interface, the server affinity type is set to ANTI_AFFINITY . If you create the anti-affintity group using the Oracle VM Web Services API, the server affinity type is set to NO_AFFINITY .

Workaround: There is no workaround for this issue.

Bug 23542910

9.6.3 Performance Impact when Less than 8 CPUs Configured for dom0

For best performance, if dom0 is configured with less than 8 CPUs, the xen_netback max_queues option in the ovs.conf file should be set to the number of CPUs configured for dom0. For example, if you configure dom0 with 6 CPUs, then you should set the value of the xen_netback max_queues option to 6.

Workaround:

  1. Start an ssh session to Oracle VM Server.

  2. Open /etc/modprobe.d/ovs.conf for editing.

  3. Set the value of the xen_netback max_queues option equal to the number of CPUs configured for dom0.

  4. Reboot Oracle VM Server for the setting to take effect.

Bug 24329130

9.6.4 Upgrading Oracle VM Server from Release 3.2.10 to 3.3.4 and then to 3.4.2 fails

When upgrading Oracle VM Server Release 3.2.10 to Release 3.4.2, and your upgrade path is from Release 3.2.10 to 3.3.4 and then to Release 3.4.2, the upgrade fails with the following message: 

kernel-uek-firmware-2.6.39-400.249.3.el5uek.noarch has missing requires of aic94xx-firmware

Workaround: After upgrading to Oracle VM Server Release 3.3.4, remove the kernel-uek-firmware-2.6.39-400.249.3.el5uek.noarch rpm package before upgrading to Release 3.4.2. For example, run the following command to delete the rpm package: 

# rpm -e --justdb kernel-uek-firmware-2.6.39-400.249.3.el5uek.noarch

Bug 24665394

9.6.5 Logical Volume Manager (LVM) is enabled on Oracle VM Server Release 3.4.1

By default, Logical Volume Manager (LVM) should not be enabled. However, in Oracle VM Server Release 3.4.1 LVM is not disabled correctly. As a result, LVM disks are inadvertently activated. This setting also applies to Oracle VM Server instances that you upgrade from Oracle VM Server Release 3.4.1.

Workaround: Disable LVM on Oracle VM Server. If you plan to upgrade to a later version, you should complete these steps before upgrading:

  1. Start an ssh session to Oracle VM Server Release 3.4.1.

  2. Open /etc/lvm/lvm.conf for editing.

  3. Locate the global { section of the file and comment out the global_filter = [ "r/.*/" ] setting option, so it is displayed as follows: 

    # For OVS, don't even open any device
# global_filter = [ "r/.*/" ]

  4. Locate the devices { section of the file. Uncomment the global_filter = [] setting and update it as follows: 

    # Since "filter" is often overridden from command line, it is not suitable
# for system-wide device filtering (udev rules, lvmetad). To hide devices
# from LVM-specific udev processing and/or from lvmetad, you need to set
# global_filter. The syntax is the same as for normal "filter"
# above. Devices that fail the global_filter are not even opened by LVM.
global_filter = [ "r/.*/" ]

  5. Save and close the file.

  6. If LVM disks are activated, you must manually deactivate them or reboot Oracle VM Server for the updates to take effect.

  7. If LVM disks are not activated, no further action is required.

Bug 24583291

9.6.6 Btrfs Root File System Not Supported on PVM Guest Virtual Machines

Note

As of Release 3.4.6, support for PVM guests is removed. For information information, see Section 8.1, “Removed Features in 3.4.6”.

Btrfs root file system is not supported on x86-based PVM guest virtual machines. For example, updating the domain type of PVHVM guests running on SUSE Linux Enterprise Server 12 SP2 or later to PVM results in the virtual machine failing to start-up. This occurs as btrfs is the default file system on SUSE Linux Enterprise Server 12 SP2 or later operating system.

Workaround: There is no workaround for this issue.

Bug 25040512

9.6.7 SUSE Linux Enterprise Server 12 SP2 and Later Not Supported For PVM Guest Virtual Machines

Although SUSE Linux Enterprise Server 12 SP2 or later is supported for both hardware virtualization (HVM) and hardware virtualization with paravirtual drivers (PVHVM), it is not possible to run this operating system as a paravirtualized domain (PVM). This is due to the default btrfs file system on SUSE Linux Enterprise Server 12 SP2 or later not being supported on PVM.

Bug 25040512

9.7 Known Issues and Workarounds

9.7.1 Cannot Connect to the Oracle VM Manager Web Interface with Microsoft Internet Explorer

In some cases, if you attempt to connect to the Oracle VM Manager Web Interface with Microsoft Internet Explorer version 10 or lower, the Oracle VM Manager Web Interface does not respond when you enter login credentials. No messages are written to the AdminServer.log file to indicate errors or provide debugging information.

Workaround: Do one of the following:

  • Reset Internet Explorer settings.

  • Use Internet Explorer version 11 or a different browser, such as Mozilla Firefox, to connect to the Oracle VM Manager Web Interface.

Bug 20691432

9.7.2 The Virtual Machine Console Window in the Oracle VM Manager Web Interface Opens in the Background when using Microsoft Internet Explorer

The virtual machine VNC or serial console windows, that can be opened from within the Oracle VM Manager Web Interface, may initially open in the background and not have immediate focus when using Microsoft Internet Explorer. This means that after you click on the menu option to open a console dialog, it initially appears as if nothing has happened as the main browser window remains in the foreground. However, the console window has opened but is in the background. This is the result of a bug in the underlying framework that is used for the development of the Oracle VM Manager Web Interface.

Workaround: Do one of the following:

  • Check in the Windows task bar to determine whether another window has been opened in the background, or use the Alt-Tab key combination to cycle through open windows.

  • Use a different browser, such as Mozilla Firefox or Chrome to connect to the Oracle VM Manager Web Interface.

Bug 21235816

9.7.3 Inconsistent Value for SPARC Domain Type

In the Oracle VM Manager Web Interface, you can select OVM/SPARC as a value for the Domain Type option used to create or edit a virtual machine. In the Oracle VM Manager Command Line Interface, and in the Oracle VM Web Services API, the value for the DomainType option can be set as LDOMS_PVM. These two values should be understood to be equivalent and refer to the hypervisor that is available on an Oracle VM Server for SPARC.

You are unable to use these values interchangeably and must use the correct value for the interface that you are using to configure the virtual machine. Check the appropriate documentation if you are unsure of the values that can be set for this option.

Workaround: There is no workaround for this issue.

Bug 20786024

9.7.4 Integer Only Values for the statValue Option

In the Oracle VM Manager Command Line Interface, you must use only integers to specify values for the statValue option when using the equalityType option for the getStatListByQuery command. Floats and decimals are not supported values for the statValue option.

Workaround: There is no workaround for this issue.

Bug 21046910

9.7.5 Aborting a Server Discovery Operation Results in Server Listed with an Unknown Operational State

Aborting a server discovery operation within Oracle VM Manager can result in the server being listed within the Oracle VM Manager Web Interface with an unknown operational state. Equally, the server object may appear as locked within the Oracle VM Manager Web Interface. This is because events triggered by the discovery operation may cause the state of the Oracle VM Manager Web Interface to fall out-of-sync with the actual state of the environment.

Workaround: To correct the operational state returned by Oracle VM Manager, rediscover any Oracle VM Servers that are affected. If you wish to remove these servers, delete them after you have performed the discovery operation.

Bug 21495833, 21495187

9.7.6 Automatic Discovery Operations can Conflict with Server Configuration Operations

In certain rare cases, an automatic server rediscovery is triggered by an error condition at the same time that a server configuration operation is taking place. In this situation, it is possible that this conflict can result in configuration information specific to the Oracle VM Server falling out-of-sync with the information reflected by Oracle VM Manager.

Workaround: To correct the information returned by Oracle VM Manager, rediscover any Oracle VM Servers that are affected.

Bug 21341236

9.7.7 Solaris 11 x86 Virtual Machine Disk Type Can Cause Kernel Panic on Boot

On Oracle VM, Solaris 11 x86 virtual machines are all installed and run with Xen Virtual Disk (XVD) disks by default. Exporting a Solaris 11 x86 virtual machine as a virtual appliance (OVA), and then importing it to a virtualization platform that supports a different disk type or disk controller, changes the path to the boot disk for the virtual machine. As a result, a kernel panic occurs for the Solaris 11 x86 virtual machine because it cannot locate the boot disk.

Workaround: Use the Solaris 11 x86 live installation ISO to boot the Solaris virtual machine. When the live installation GUI displays, open a terminal window and change to the root user. Run zpool import -f rpool and then run shutdown. After the virtual machine shuts down, remove the live installation ISO and boot from the disk.

Bug 21253581

9.7.8 Solaris 10 x86 Virtual Machine Installation Fails To Boot

On Oracle VM, virtual machines that are installed with Solaris 10 directly from the installation ISO may fail to boot immediately after installation and can return the error: 'Bad PBR Sig'.

This bug is related to the way in which virtual disks are referenced in the virtual machine configuration file, where they are configured as Xen Virtual Disk (XVD) disks by default. For Solaris to recognize this disk type, a patch needs to be applied to the Solaris installer to recognize the new disk type and to allow it to write to the boot block of the disk. This patch can been applied for network installs, but is not applied to the installation ISO, therefore additional steps must be taken when performing an installation from the ISO.

Workaround: The workaround provided here has been tested for Solaris 10u11. If you require that you complete an installation of Solaris 10u11 from ISO, it is possible to rectify the issue described here by performing the following steps:

  • Reboot the virtual machine and use the ISO to perform the boot.

  • During the boot process select the option to go to command line mode (option 6).

  • At the command prompt, run installgrub to manually install the grub loader to the boot partition, as follows: 

    /a/sbin/installgrub -m /a/boot/grub/stage1 /a/boot/grub/stage2 /dev/rdsk/c1t0d0s0

    Substitute c1t0d0s0 with your device as appropriate.

  • Reboot the virtual machine and the installation completes.

Solaris 11 is unaffected by this issue.

Bug 21775043

9.7.9 Oracle Linux 7 Virtual Machines Imported from a VirtualBox Virtual Appliance Do Not Start on Oracle VM

Virtual machines running Oracle Linux 7 exported from Oracle VirtualBox as a virtual appliance and then imported into Oracle VM may not boot correctly and may exit to an emergency shell. This is caused by a missing xen-blkfront driver in the initramfs image. Typically, output during boot, for affected systems, appears as follows: 

[ OK ] Reached target Basic System.
dracut-initqueue[227]: Warning: Counld not boot.
dracut-initqueue[227]: Warning: /dev/mapper/ol-root does not exist
dracut-initqueue[227]: Warning: /dev/ol/root does not exist
dracut-initqueue[227]: Warning: /dev/ol/swap does not exist
             Starting Dracut Emergency Shell ...
Warning: /dev/mapper/ol-root does not exist
Warning: /dev/ol/root does not exist
Warning: /dev/ol/swap does not exist

Workaround: There are two workarounds to this problem. The first involves adding the missing drivers prior to exporting the Oracle Linux 7 virtual machine from Oracle VirtualBox. To do this, run the following command as root before you perform the export: 

# dracut --add-drivers "xen-blkfront xen-netfront" --force

If you are unable to perform this step prior to the export, you can temporarily boot the virtual machine as an HVM and add the following boot option in GRUB prior to boot: 

xen_emul_unplug=never

Once the virtual machine has booted, you can add the missing drivers by running the following command as root: 

# dracut --add-drivers "xen-blkfront xen-netfront" --force

Reboot the virtual machine once the drivers have been added to the initramfs.

Bug 21244825

9.7.10 (x86 Only) IOStat Stays at 100% Util for Oracle Linux Virtual Machines After a Storage Migration

When a virtual machine running a kernel prior to UEK 3.8.13-68.2.1 is migrated from a local repository using the 'live migrate with storage' option, and all IO has been terminated, the IOStat on the virtual machine remains stuck at 100% Util.

This is related to a kernel bug where a call to the Xen blkfront driver caused the IO statistics to not be properly refreshed after a migration was completed, causing the accounting to fall out of sync with actual IO activity.

Workaround: This bug is resolved in UEK 3.8.13-68.2.1 and greater. To avoid this issue, upgrade the kernel used by your virtual machine.

Bug 21614184

9.7.11 Serial Console Does Not Respond for Oracle Linux 7 HVM Virtual Machines

The serial console does not respond, or provide any output, for virtual machines based on Oracle Linux 7 that are of domain type HVM or PVHVM.

Workaround: Configure the GRUB bootloader for the virtual machine as follows:

  1. Add the following lines to the /etc/default/grub file: 

    GRUB_CMDLINE_LINUX=" console=ttyS0 console=tty0 crashkernel=auto \
  rd.lvm.lv=ol/root rd.lvm.lv=ol/swap rhgb quiet"
GRUB_TERMINAL="serial console"
GRUB_SERIAL_COMMAND="serial --unit=0"

  2. Run the following command to update the grub.cfg file: 

    grub2-mkconfig -o /boot/grub2/grub.cfg

  3. Restart the virtual machine.

Note that the preceding lines provide suggested configuration. You should consult the GNU GRUB manual for descriptions of each line and additional configuration options to suit your business needs. The GNU GRUB manual is available from the following directory on the virtual machine: usr/share/doc/. Additionally, you should refer to the About the GRUB2 Boot Loader section of the Oracle Linux Administrator's Guide for more information.

Tip

Use the following keys to navigate the boot menu: v for down-arrow and ^ for up-arrow.

Bug 21229730

9.7.12 Error When Attaching CD-ROM Drives to Running HVM Virtual Machines

If you attempt to attach a CD-ROM drive to an available slot on a running virtual machine of domain type HVM, an error occurs and it is no longer possible to use the specified slot. For example, after you attempt to attach a CD-ROM drive to slot 2, you then attempt to attach a virtual disk to slot 2. In both cases an error occurs and slot 2 becomes unusable.

This issue occurs because running HVM virtual machines do not support the function to attach CD-ROM drives.

Workaround: Stop the virtual machine and then attach the CD-ROM drive.

Bug 21300944

9.7.13 Deleting Bond Ports for VLANIF Interfaces Deletes the VLANIF

Deleting a bond port that is in use by a VLANIF interface configuration also deletes the VLANIF interface.

Workaround: There is no workaround for this issue.

Bug 21217364

9.7.14 (SPARC Only) NFSv4 Delegation Interferes with Cluster Heartbeat, Using NFSv3 Results in RPC Error PROGRAM/VERSION MISMATCH

Server pool clustering on SPARC based server pools can cause servers to automatically fence due to NFSv4 write delegation revocation timeouts which exceed the timeout value for the cluster.

It is not possible to use NFSv3 as this results in an RPC error reporting a "PROGRAM/VERSION MISMATCH" and can cause the cluster database to hang due to the NFSv3 locking mechanism. This is expected behavior.

Workaround: There is no workaround for this issue. Only NFSv4 is supported for SPARC clustering, and you must disable delegation for this to work.

Bugs 18997487, 20443098, 21317571

9.7.15 Cannot Import OVF Files into Oracle VirtualBox

You cannot successfully import a virtual appliance into Oracle VirtualBox if you exported that virtual appliance from Oracle VM as an extracted OVF that includes three separate files, a package.ovf file, a package.mf file, and a disk.vmdk file.

Workaround: Delete the package.mf file before you import the virtual appliance into Oracle VirtualBox. Alternatively, export the virtual appliance from Oracle VM as an OVA file.

Bug 21261077

9.7.16 Oracle VM Manager Cannot Receive Messages from SPARC Virtual Machines

After sending messages to a virtual machine running on Oracle VM Server for SPARC to configure network settings, Oracle VM Manager does not receive messages that indicate that the network was configured successfully.

Workaround: There is no workaround for this issue.

Bug 21062991

9.7.17 Cannot Boot Oracle VM Server Installed on an HP DL380 Gen9 Server in UEFI mode

In some cases, after you install Oracle VM Server on an HP DL380 Gen9 server in UEFI mode, you cannot successfully boot Oracle VM Server. This is the result of an issue with HP Smart Array firmware versions lower than 2.52.

Workaround: Update the HP Smart Array firmware to version 2.52 or higher.

Bug 21466879

9.7.18 (x86 Only) Cannot Boot Oracle VM Server with Broadcom 57840S CNA Card After PXE Install

If you perform a network installation of Oracle VM Server on an x86 system using the Broadcom 57840S CNA Card, you are unable to boot using FCoE. This issue is the result of a bug in the anaconda installer, where the installer is unable to see all of the disks when the system is booted using PXE.

Workaround: Do not use PXE boot to install systems that make use of this hardware.

Bug 20830496

9.7.19 Changing to Dynamic Mode IP Address Assignment Does Not Rollback if the Server is Unable to Obtain an IP Address

If you choose to change the IP address assignment mode to 'Dynamic' for any Oracle VM Server network interface or port within Oracle VM Manager, the operation that is triggered is a multi-step operation. First, the network configuration is changed on the server and within the Oracle VM Manager database. Then the server attempts to obtain an IP address from the DHCP server. If this second operation fails, the server returns an error but the Oracle VM Manager is unable to rollback the initial operation. The result is that the server remains configured for DHCP but the port does not have an IP address assigned.

If the port is used for the Management Network, it is possible that the server is no longer accessible to the Oracle VM Manager and you must correct the networking manually on the server. This may require you to reissue a DHCP request by running the dhclient command on the command line of the server, or it may require that you reconfigure the network interface with a static IP address and network mask. You may do this temporarily using the ip addr command on the command line. Once your server has an IP address within the management network range, you must rediscover the server.

Workaround: There is no workaround for this issue. You must ensure that your infrastructure is fully functional and that servers are able to obtain IP addresses from the DHCP server if you choose to use Dynamic Mode for IP address assignment for a server. Alternately, use Static IP addressing.

Bug 21370280

9.7.20 Drag and Drop from Blank Space of Object

It is not possible to drag and drop one or more virtual machines or Oracle VM Servers if you select the text area of the object or the expansion arrow next to the object.

Workaround: Select an area of blank space in the object to drag and drop that object.

Bug 21687716

9.7.21 Error When Using the Administrator Tool in Parallel

If two or more users connect to the same instance of Oracle VM Manager and simultaneously attempt to create users with the Oracle VM Manager Administrator Tool (ovm_admin), an error occurs and the following message is written to AdminServer.log: 

Problem invoking WLST - Traceback (innermost last): File
"/u01/app/oracle/ovm-manager-3/ovm_wlst/jython/ovm/admin/createuser.py"

Workaround: Ensure that only one user at a time connects to Oracle VM Manager to use the Oracle VM Manager Administrator Tool (ovm_admin).

Bug 21679271

9.7.22 Oracle VM Manager Command Line Interface TCP Port Conflicts With Port Reserved for NDMP

The Oracle VM Manager Command Line Interface uses TCP port 10000 by default. This port is usually reserved for the Network Data Management Protocol (NDMP).

Workaround: Change the default port for the Oracle VM Manager Command Line Interface if you require TCP port 10000 for other applications that use NDMP, see SSH Port in the Oracle VM Manager Command Line Interface User's Guide.

Bug 21960034

9.7.23 When A Virtual Machine Using A Shared Virtual Disk Is Moved To Another Repository, The Shared Virtual Disk Is Cloned

When a virtual machine that is using a shared virtual disk is moved to another repository, the shared virtual disk is cloned during the move process to allow the virtual machine to continue to use the resources on the shared virtual disk. This may not be the behavior that you expect when you move a virtual machine and could result in duplicated data resources that could eventually become unsynchronized and difficult to reconcile.

This design decision allows virtual machines to be restarted after a move with minimum impact as all used resources continue to be available. However, the behavior is not intuitive and may be modified in subsequent releases.

Workaround: Be aware of the existing behavior and detach any shared disks before moving the virtual machine. If the virtual machine needs to make use of resources on the shared disk, detach the shared disk from all virtual machines and move it separately before reconfiguring your virtual machine to use it again. Make sure that all virtual machines that require access to the same shared disk are located in the same repository.

Bug 21316001

9.7.24 Unable to Remove a Storage Array if a File System Exists on a LUN

In the case where a repository is hosted on a LUN on a storage array and ownership of the repository is released, a rule violation is reported if you attempt to delete the storage array. This issue only affects storage arrays. File servers are unaffected.

Workaround: To resolve this issue you must completely unconfigure the elements of the storage array before attempting to remove it. Remove ownership of any repositories hosted on the storage array, remove the storage server from any access groups and finally delete the storage array.

Bug 22313650

9.7.25 Cluster Failure After An Offline Server is Removed From the Cluster While Another Cluster Member is Offline

If more than one Oracle VM Server within a clustered server pool is offline at the same time, and one of those servers is removed from the server pool, when the other offline servers come back online, the cluster remains in a failed state. This is because the cluster configuration falls out of sync across the members of the cluster.

Whenever a server is added or removed from a cluster, Oracle VM Manager triggers an operation on each Oracle VM Server in the cluster to update the cluster configuration information. However, if any servers are offline at this time, they are unable to receive the updated configuration resulting in a configuration mismatch between the cluster configuration on the server and the actual configuration for the rest of the cluster. As a result, the server is no longer able to participate in the cluster.

On an x86 platform, this situation is simply represented within Oracle VM Manager and can be easily resolved from within Oracle VM Manager. On a SPARC platform, this can cause a server to repeatedly reboot when it comes online again due to a cluster panic.

Workaround (SPARC): In the case where a SPARC server is rebooting continuously, the problem is caused by the server attempting to rejoin the cluster even though it is no longer a member of the cluster. As a result, the server can panic repeatedly with a message similar to the following: 

panic[cpu5]/thread=2a102a13c60:
 **** dlm FENCING this system by PANICing

If this issue occurs and a server is repeatedly panicking then you can prevent the server from panicking again by stopping the ovs-agent service and deconfiguring the cluster. To do so, connect to the server as root and run the following commands: 

# svcadm disable -s ovs-agent
# dlmcconf -S

If you are unable to run these commands because the server is panicking too quickly, then boot the server in single-user mode: 

ok boot -s

Disable the ovs-agent service: 

# svcadm disable ovs-agent

Reboot the server. The server stops panicking while the ovs-agent service is disabled. If you want to re-enable the ovs-agent service, then you must resolve the cluster configuration issue first.

Once you have resolved the cluster configuration issue, you can acknowledge the server cluster failure event within Oracle VM Manager to resume normal operations.

Workaround (x86): To restore the environment to normal operation, you must first acknowledge the server cluster failure event within Oracle VM Manager. Remove any servers from the server pool that were offline when you made the configuration change, and then add them back to the server pool, so that the cluster configuration information can be properly refreshed for that server.

Bug 22304185, 18776654

9.7.26 LSI Logic / Symbios Logic MegaRAID SAS 1078 is not supported on Oracle VM Server

Oracle VM Server may not be able to detect disks on systems using the Dell PowerEdge Expandable RAID Controller 5 which uses LSI Logic/Symbios Logic MegaRAID SAS 1078. This is due to the unavailability of compatible driver firmware to enable device support within UEK4 from the hardware manufacturer. You may not be able to successfully upgrade servers using this hardware to Oracle VM Server 3.4.1 or above.

There are no workarounds for this issue. You must upgrade your hardware.

Bug 21843131, 21223382

9.7.27 (SPARC Only) Hot-added CD-ROM Device is Detected As Hard Disk

If you hot add a virtual CD-ROM device to a virtual machine running on Oracle VM Server for SPARC, the CD-ROM device is not detected as a removable device and is not shown when you run the rmformat or format commands. Instead, the device is listed as a virtual disk and is mapped this way as a device alias. This can be verified using the OpenBoot PROM to run the show-disks and devalias commands.

Workaround: There is no workaround to this issue. Remain aware that when hot-adding a virtual CD-ROM device, this is detected as a virtual disk instead and is mapped in Solaris as a /dev/dsk/c1dXsX entry.

Bug 21278127

9.7.28 Fibre Channel LUNs Missing From Oracle VM Manager After Refresh

In some instances, after you refresh physical disks in the Storage tab in Oracle VM Manager Web Interface, fibre channel LUNs might not be included in the refresh. As a result, the LUNs appear to be missing even though they are online at the server.

Workaround: You should first examine all components of the infrastructure, such as switches, cables, and HBAs, to ensure that no defects or failures exist. If, after a subsequent refresh, the LUNs still appear to be missing, you can perform the following steps to resolve this issue:

  1. List all attached SCSI devices. 

    # lsscsi
[0:0:0:0]    disk    HP       LOGICAL VOLUME   1.34  /dev/sda
[0:0:0:1]    disk    HP       LOGICAL VOLUME   1.34  /dev/sdb
[0:3:0:0]    storage HP       P440ar           1.34  -
[1:0:0:0]    disk    SUN      ZFS Storage 7330 1.0   /dev/sdc
[1:0:0:1]    disk    SUN      ZFS Storage 7330 1.0   /dev/sdd
[1:0:0:2]    disk    SUN      ZFS Storage 7330 1.0   /dev/sde
[1:0:1:0]    disk    SUN      ZFS Storage 7330 1.0   /dev/sdf
[1:0:1:1]    disk    SUN      ZFS Storage 7330 1.0   /dev/sdg
[1:0:1:2]    disk    SUN      ZFS Storage 7330 1.0   /dev/sdh
[1:0:2:1]    disk    NETAPP   LUN              7330  /dev/sdo
[1:0:2:2]    disk    NETAPP   LUN              7330  /dev/sdp
[1:0:2:3]    disk    NETAPP   LUN              7330  /dev/sdq

  2. Change to the SCSI address of the affected LUNs.

    Based on the preceding example, you can enter rport-1:0-2 to cover [1:0:2:1], [1:0:2:2], [1:0:2:3], and so on. 

    # cd /sys/class/fc_remote_ports/rport-1:0-2/

  3. Check the timeout value of the fast_io_fail_tmo option. 

    # cat fast_io_fail_tmo
5

  4. Increase the timeout value to just over the amount of time it takes for the LUN to return to the running state. The value that you set should be less than the value of the dev_loss_tmo option in the same directory. You should need to change only one of the affected LUNs. 

    # echo 15 > fast_io_fail_tmo

  5. Add the command to /etc/rc.local to make the change permanent. 

    # cat /etc/rc.local
#!/bin/sh
#
# This script will be executed *after* all the other init scripts.
# You can put your own initialization stuff in here if you don't
# want to do the full Sys V style init stuff.

echo 15 > /sys/class/fc_remote_ports/rport-1:0-2/fast_io_fail_tmo

Bug 22322461

9.7.29 (SPARC Only) Size of Physical Disk in Oracle VM Manager Incorrect When LUN Size is Decreased on Storage Array

In the event that a LUN is resized directly on the storage array and that LUN is used by an Oracle VM Server for SPARC, the size of the LUN may not be correctly updated in Oracle VM Manager if the LUN size was decreased. This is due to the mechanism that the Oracle VM Agent for SPARC uses to obtain and report on the LUN size, which does not update once the disk is mapped on the server.

This problem does not occur when the LUN size is increased.

Workaround: The reported LUN size can be updated by rebooting the Oracle VM Server for SPARC where the LUN is mapped as a physical disk after you have resized the LUN. Once the server is back online, you must refresh the storage array where the affected LUN is located. The correct size is displayed for the LUN within Oracle VM Manager.

Bug 22087392

9.7.30 The Apparent Size of a Repository in Oracle VM Manager Web Interface Does Not Update Automatically

The apparent size of a repository filesystem is displayed in the Oracle VM Manager Web Interface on the Repositories tab in the Info perspective for a selected repository. This information is a statistical value that is updated periodically within Oracle VM Manager but which is not updated automatically within the table within the Oracle VM Manager Web Interface. The information displayed for this value matches the information available within Oracle VM Manager at the time that the page loads.

Workaround: To refresh the information displayed for the apparent size of a repository within the Oracle VM Manager Web Interface, you may click on an alternate tab and then return to the view of the repository to force the page to refresh.

Bug 22147488

9.7.31 Virtual Disks Do Not Display on Solaris 10 Guests

If you add virtual disks to a Solaris 10 virtual machine that is stopped, those virtual disks do not display when you list the virtual disks for the virtual machine.

Workaround: This is a Solaris 10 issue that you can resolve as follows:

  1. Start the virtual machine.

  2. Run the following command on the virtual machine: devfsadm -v

Bug 22460766

9.7.32 Oracle Linux 5 Virtual Machines Converted from Physical Hosts Do Not Boot

In some cases, virtual machines based on Oracle Linux 5 do not boot when started. This can occur if you used the P2V utility to convert a physical host based on Oracle Linux 5 to a virtual machine and then imported that virtual machine to Oracle VM Manager.

Unbreakable Enterprise Kernel version kernel-uek-2.6.32-100.34.1 adds support for paravirtualized drivers in a HVM guest on Oracle VM. Starting with this kernel version, the default is to present only paravirtualized drivers when running in a hardware virtualized guest. To run kernel-uek, including the drivers, fully hardware virtualized, an additional kernel boot parameter xen_emul_unplug=never must be added to the boot parameters in /etc/grub.conf.

Workaround: Add the kernel boot parameter as follows: 

kernel /vmlinuz-2.6.32-100.34.1.el5uek ro root=/dev/VolGroup00/LogVol00  xen_emul_unplug=never

Bug 22285006

9.7.33 SSL Exception in AdminServer Log File

In rare cases, the following SSL exception is written to AdminServer.log after attempting to perform an operation on a virtual machine: 

<Error> <ovm.mgr.ras.WebSocketHandler>
<SSLException, writeNWBuf: java.nio.HeapByteBuffer[pos=0 lim=27 cap=16971] is
not large enough for SSLEngine.unwrapAndHandleResults, results: Status =
BUFFER_OVERFLOW HandshakeStatus = NOT_HANDSHAKING
bytesConsumed = 0 bytesProduced = 0
java.io.IOException: SSLException, writeNWBuf: java.nio.HeapByteBuffer[pos=0
lim=27 cap=16971] is not large enough for SSLEngine.unwrapAndHandleResults,
results: Status = BUFFER_OVERFLOW HandshakeStatus = NOT_HANDSHAKING
bytesConsumed = 0 bytesProduced = 0

Workaround: This exception is due to a known issue with Oracle WebLogic Server. You can ignore the exception. Subsequent attempts to edit or perform an operation on the virtual machine should be successful.

Bug 22451280

9.7.34 Oracle VM Servers in a Server Pool Must Be at the Same Minor Version

Each instance of Oracle VM Server in a server pool must be at the same minor version. In other words, all servers in a server pool must be at Release 3.2.x, or Release 3.3.x, or Release 3.4.x. For example, a server pool cannot contain one or more instances of Oracle VM Server Release 3.3.1 and one or more instances of Oracle VM Server Release 3.2.10.

During an upgrade, each instance of Oracle VM Server in the server pool is upgraded separately, which results in a mixed version server pool until the upgrade process is complete. For this reason, mixed version server pools are supported only during an upgrade of Oracle VM Server.

Workaround: There is no workaround for this issue.

Bug 22516403

9.7.35 Oracle VM Manager Displays Incorrect Details for ZFS Storage Appliances After Cluster Failover

In environments where two or more ZFS storage appliances are in an active/active cluster configuration, and a cluster failover occurs, it is possible for Oracle VM Manager to display incorrect IP addresses and file system information.

Workaround: Stop all running virtual machines, unpresent the repositories, and then delete the NFS servers from Oracle VM Manager.

Bug 21466858

9.7.36 Oracle VM Server Upgrade Not Possible with NetApp Storage plug-in

It is not possible to upgrade Oracle VM Server from Release 3.3.x to Release 3.4.1 if the NetApp Storage plug-in, netapp-osc-linux, is installed.

Workaround: Remove the netapp-osc-linux plug-in before you upgrade Oracle VM Server to Release 3.4.1. When the upgrade successfully completes, reinstall the netapp-osc-linux plug-in.

Bug 21052337

9.7.37 Oracle VM Manager Web Interface Incorrectly Displays Oracle VM Server Upgrade Required

The Update Required column incorrectly indicates that an update is required for Oracle VM Server when you create a server update repository but do not enable that repository and then refresh the servers to which that server update repository applies. If you subsequently attempt to upgrade Oracle VM Server, the upgrade does not occur.

Note

The Update Required column is located in the Servers perspective in the management pane on the Servers and VMs tab.

Workaround: There is no workaround for this issue.

Bug 22374697

9.7.38 The Oracle VM Manager Silent Install Option Using A YAML Configuration File Is Deprecated

The function to install Oracle VM Manager silently using a predefined YAML formatted configuration file is intended only for internal use during testing at Oracle. The option to install using this facility is deprecated for customer use and is undocumented in this release.

Workaround: There is no workaround for this issue.

Bug 22478672

9.7.39 Aborting An Oracle VM Server Upgrade Initiated By The UpgradeServers.py Script May Result In Inconsistent Reporting on Upgrade Status

In the unlikely case that an Oracle VM Server is in the process of being upgraded using the UpgradeServers.py Script and the upgrade job is aborted from within Oracle VM Manager, the upgrade on the Oracle VM Server may complete but the UpgradeServers.py script may also report that the upgrade has failed. Equally, Oracle VM Manager may report an incorrect version or upgrade status for the Oracle VM Server.

Workaround: Reboot the affected server and then disable maintenance mode on the server.

Bug 22488078

9.7.40 Kdump On Oracle VM Server Only Supports Dump Levels 0 and 1

The makedumpfile command, used to manually trigger a vmcore dump file is documented to support dump levels from 0 to 31 using the -d switch. Notably, the manual page states that it is possible to specify a value higher than 1 on an x86_64 machine running the Xen hypervisor. This is not the case for Oracle VM Server. Only the values 0 and 1 can be specified for this parameter, or a vmcore file is not generated.

Workaround: There is no workaround. Only use dump levels of 0 or 1 if using the makedumpfile command.

Bug 22329501

9.7.41 Network Receive Statistics For A Virtual Machine Are Invalid And Set To -1

DRS/DPM algorithms that monitor network load on a virtual machine do comparisons between the most recent network statistics and the previous network statistics to determine the deltas that can then be evaluated in terms of user-defined thresholds.

The network statistics are captured as unsigned integers. This means that they have a range of [0, 4294967295]. If the number of bytes received is greater than 4 GB the value of the integer overflows and begins from 0 again. In this situation, it is possible the start value for the statistic recorded by Oracle VM Manager is greater than the end statistic. Once this has occurred, Oracle VM Manager sets this statistic to -1 as it is no longer meaningful or valid.

If you have a DRS/DPM policy set to trigger on network load, the following error may appear in the log: 

VM rx stats are not valid, the stats wrapped around.

In the event that an invalid network statistic is returned, the statistic is ignored by the DRS/DPM algorithm for that cycle. Normal behavior resumes on the next cycle. This bug has a very limited effect and is only noticeable on very loaded networks where the occasional log message may appear.

Workaround: There is no workaround required for this issue. It is expected behavior and the issue is resolved automatically by the environment.

Bug 22550810

9.7.42 CPU Affinity After Live Migration of Virtual Machine on Non-Uniform Memory Access (NUMA) Systems

After a live migration of a virtual machine between two servers with NUMA architecture, the CPU affinity is set to "any CPU". As a result, the migrated virtual machine uses memory across multiple nodes, which degrades performance of the virtual machine.

Workaround: Before performing the live migration of virtual machines, configure CPU pinning in the guest vm.cfg file. For example, to pin a virtual machine with 4 VCPUs to physical CPUs 4, 5, 6, and 7, add the following line to the vm.cfg: 

cpu = "4-7"

Bug 22382809

9.7.43 Virtual Machines Based on Oracle Solaris 11 Do Not Boot

In some cases virtual machines based on Oracle Solaris 11 lower than 11.3 do not boot.

Workaround: Upgrade the guest operating system to Oracle Solaris 11.3 or higher.

Bug 22085158

9.7.44 Changing Domain Type from Xen HVM to Xen HVM PV Drivers Causes Informational Messages in AdminServer.log

Changing the domain type of a virtual machine from Xen HVM to Xen HVM PV Drivers results in the following informational messages of type vm.domain.mismatch and vm.os.mismatch being written to AdminServer.log.

This issue occurs because Oracle VM Server does not distinguish between the Xen HVM and Xen HVM PV Drivers domain types. When the domain type of the virtual machine changes to Xen HVM PV Drivers, Oracle VM Server reports the domain type of the virtual machine as Xen HVM to Oracle VM Manager. As a result, Oracle VM Manager falsely detects an error and logs the informational messages.

Workaround: You can ignore the informational messages and change the domain type of the virtual machine to Xen HVM PV Drivers. The virtual machine successfully loads the PV drivers despite these messages.

Bug 21163194

9.7.45 Single-Stream TCP Throughput is Degraded for Guest Virtual Machines Running UEK Release 4

The single-stream TCP throughput in the TX subgroup is degraded for guest virtual machines running Oracle Unbreakable Enterprise Kernel (UEK) Release 4.

Workaround: This issue occurs because UEK Release 4 includes support for the TCP Small Queues (TSQ) mechanism. To resolve this issue, you can configure the guest virtual machine with the following command: 

echo 1048576 > /proc/sys/net/ipv4/tcp_limit_output_bytes

Bug 22546060

9.7.46 Error Partitioning Caused by Disk Label or Format in UEFI Mode

If you attempt to install Oracle VM Server in UEFI mode on a virtual drive that is not GPT or only MBR, a partitioning error occurs in the Anaconda installation program.

Workaround: Do the following to work around this issue:

  1. Select OK to return to the previous installer screen.

  2. Enter Alt+F2 to open a shell prompt.

  3. Use the parted command to change the label of the disk. Alternatively, use the dd command to clean up the disk.

  4. Return to the installation program and then select Back to return to the previous installer screen.

  5. Ensure that storage is discovered and then continue the installation.

Bug 22503847

9.7.47 Kernel Panic when Booting Guest Virtual Machine with Oracle Solaris 11.2

When creating a virtual machine with a guest operating system of Oracle Solaris 11.2 on Oracle VM Server, a kernel panic occurs when the virtual machine boots.

Workaround: Upgrade the guest operating system to Oracle Solaris 11.3.

Bug 22457279

9.7.48 Oracle VM Manager Web Interface Goes to Blank Screen with Mozilla Firefox on a Linux System

When creating a new repository, the Oracle VM Manager Web Interface temporarily displays a blank screen and the following message is written to AdminServer.log: 

<Error>
<oracle.adfinternal.controller.application.AdfcExceptionHandler> <ADFC-50018>
<ADFc: No exception handler was found for an application exception.
javax.faces.FacesException: Broken pipe

This issue occurs when using a version of Mozilla Firefox that is lower than Version 44.0 on a Linux system to access the Oracle VM Manager Web Interface. This issue does not affect the jobs in Oracle VM Manager to create a new repository.

Workaround: Upgrade to Mozilla Firefox Version 44.0.

Bug 22553553

9.7.49 HugePages Deprecated for PVM Guests as of Oracle VM Release 3.4.1

The HugePages feature is deprecated in Oracle VM Release 3.4.1. You should not enable HugePages when creating or editing virtual machines with a domain type of PVM in the Oracle VM Manager Web Interface or Oracle VM Manager Command Line Interface. This feature will be removed in a future release of Oracle VM.

If you have HugePages enabled for any PVM guests, Oracle recommends that you change the domain type for virtual machines from Paravirtualized (PVM) to Hardware virtualized, with paravirtualized drivers (PVHVM). If you cannot change the domain type for a virtual machine, you should disable the HugePages setting and then restart the virtual machine.

Workaround: There is no workaround for this issue.

Bug 22540900

9.7.50 HugePages Does Not Display as Selected in Oracle VM Manager Web Interface

The Enable Huge Pages option in the Edit Virtual Machine dialog box does not display as selected when virtual machines with a PVM domain type are running.

Workaround: There is no workaround for this issue. However, Oracle recommends that you do not enable HugePages for any PVM guests. See Section 9.7.49, “HugePages Deprecated for PVM Guests as of Oracle VM Release 3.4.1”.

Bug 22746016

9.7.51 Kdump Service Hangs at Discovery of FCoE Device with Broadcom CNA on HP Servers

If you install Oracle VM Server in Fiber Channel over Ethernet (FCoE) LUNs over Broadcom Converged Network Adapters (CNA) on HP servers, the kdump service hangs when discovering FCoE LUNs.

Workaround: There is no workaround for this issue.

Bug 22508971

9.7.52 Do Not Discover Oracle VM Servers Earlier than Release 3.2.10

If you attempt to discover an instance of Oracle VM Server that is earlier than Release 3.2.10 with Oracle VM Manager Release 3.4, the following occurs:

  • The discovery job fails with the following error message: 

    OVMAPI_4010E Attempt to send command: update_core_api_bindings to server:
server_fqdn failed. OVMAPI_4004E Sync command failed on
server: server_ip. Command: update_core_api_bindings, Server error:
org.apache.xmlrpc.XmlRpcException: urlgrabber.grabber.URLGrabError:[Errno 14]
problem making ssl connection

  • Conflicts occur between the databases that Oracle VM Manager requires for maintaining synchronization between internal components. As a result, the Oracle VM Manager Web Interface can display inaccurate information about the Oracle VM environment.

  • The instances of Oracle VM Server, earlier than Release 3.2.10, that you discover cannot be removed with either the Oracle VM Manager Web Interface or the Oracle VM Manager Command Line Interface.

Workaround: Do not use Oracle VM Manager Release 3.4 to discover or interact with any instances of Oracle VM Server earlier than Release 3.2.10.

Bug 22735203

9.7.53 (SPARC Only) Cannot Export Virtual Appliance with Physical Disk Access

If you create a virtual machine on an Oracle VM Server for SPARC with a physical disk access, it is not possible to export that virtual machine as a virtual appliance.

In this case, when you attempt to export the virtual machine as a virtual appliance with the Oracle VM Manager Command Line Interface, the following error message is displayed: 

Server error: vbox-img didn't return expected value (0,)

    CMDERR:
      vbox-img: error: VD: error VERR_FILE_NOT_FOUND opening image file '/dev/mapper/<disk_Id>'
      vbox-img: error: Error while opening source image: File not found.
(VERR_FILE_NOT_FOUND)

Workaround: There is no workaround for this issue.

Bug 22745868

9.7.54 (SPARC Only) MTU Resets to the Default of 1500 After Server Power Off

If you configure the MTU parameter of a network interface for an ethernet bridge and then power off the Oracle VM Server for SPARC, the MTU used by the bridge after the boot can be reset to the default value of 1500 instead of the configured value.

Workaround: When the interface used for the bridge is persistently plumbed in the primary domain, the bridge is created with the appropriate MTU parameter after the server power off and reboot. Use the ipadm(1M) command to persistently plumb a network interface, as in the following example: 

# ipadm create-ip net0

Bug 22705349

9.7.55 Oracle VM Server Installer Fails When Modifying Swap Partitions

An exception occurs and the Oracle VM Server installer stops abruptly if you select File System Options while editing a swap partition from the Partitioning Type screen.

Workaround: Do one of the following from the Review Partition Layout screen:

  • Select Yes to review and modify the partitioning layout. On the Partitioning screen, delete the swap and then create a new one.

  • Select No to skip the partitioning layout check and proceed with the installation.

Bug 22807435

9.7.56 Kernel Startup is Slow after Upgrading to UEK4 on PVHVM Guests

If you upgrade the kernel on a PVHVM virtual machine to UEK4, the kernel init process takes several minutes to complete.

Workaround: Upgrade GRUB to the most recent version.

Bug 22761370

9.7.57 Cannot Enable High Availability (HA) for Virtual Machines with HugePages

If you attempt to enable HA for a running virtual machine that has HugePages enabled, the job to enable HA fails with the following error message: 

OVMRU_005002E Operation not allowed- Virtual Machine: vm_name, is not STOPPED. It is RUNNING.

This error occurs if you attempt to enable HA in either the Oracle VM Manager Web Interface or the Oracle VM Manager Command Line Interface.

The preceding error message does not accurately describe the issue. The fact that the virtual machine is in the running state does not affect the ability to enable HA. The error results from the fact that HugePages is enabled for the virtual machine.

Workaround: Disable the HugePages setting and then restart the virtual machine. After the virtual machine starts, enable HA.

Bug 22753803

9.7.58 Kdump is Not Supported on FCoE SANs

Kdump is not supported if you install Oracle VM Server on an Fiber Channel over Ethernet (FCoE) SAN .

Workaround: There is no workaround for this issue.

Bug 22816860

9.7.59 Cannot Edit Running Virtual Machines with HugePages and HA Enabled

It is not possible to edit a running virtual machine through the Oracle VM Manager Web Interface if that virtual machine has both HugePages and high availability (HA) enabled. The job to edit the virtual machine fails with the OVMRU_005002E Operation not allowed error message.

Workaround: There is no workaround for this issue.

Bug 22814818

9.7.60 Import URL Cannot Exceed Maximum Character Length

The maximum character length of the URL to import objects into Oracle VM Manager, such as virtual appliances, virtual disks, or ISO files, is 3,988. The maximum character length of the URL to import templates is 3,985.

If you attempt to import an object from a URL that exceeds the maximum character length, an error occurs and the import fails, as in the following example: 

OVMAPI_D000E Cannot set description on object: my_virtual_appliance.ova.
 Its length: 4,012, exceeds the maximum size: 4,000.

Even though the error message indicates the maximum character length is 4,000, a prefix is automatically added to the URL during the import job. As a result, the effective character limit for the URL is 3,988 or 3,985.

Workaround: There is no workaround for this issue. Ensure that the origin URL for any object that you plan to import into Oracle VM Manager is not greater than the maximum character limit.

Bug 23073177

Copyright © 2011, 2020 Oracle and/or its affiliates. Legal Notices