Bugs Affecting the Oracle VM Server for SPARC 3.6.1 Update Release

Language:

Any bugs described in this section might also affect the Oracle VM Server for SPARC 3.6 software.

Migrated Domains Do Not Work on Targeted Node in OVM 3.6.1 With PM, Named Cores

Bug ID 29349849: On the target node, migrated domains do not function. In addition, it takes about 4 minutes to execute an ldm command on the control domain.

This problem occurs when a live migration is performed in the following configuration:

Power Management (PM) function is enabled on the target node
Setting named resource for CPU cores on the source migration domain

Recovery: Execute the following commands on the target node:

# ldm stop -f <migrated-domain>
# svcadm restart ldmd
# ldm start <migrated-domain>

Workaround: Turn off the PM function on the target node before ldm migrate-domain is executed on the source node. The PM function is enabled by default on Fujitsu SPARC systems.

Choose one of the following methods as a workaround.

Set the power-saving operation to false with related SMF property before ldm migrate-domain is executed. Subsequently, set the power-saving operation to true again after migrated domain state is active.
- On an Oracle SPARC system:
  - To disable PM manually, run the following commands on the target node:
```
# svccfg -s ldmd setprop ldmd/pm_enabled=false
# svcadm refresh ldmd
# svcadm restart ldmd
```
  - To enable PM manually, run the following commands on the target node:
```
# svccfg -s ldmd setprop ldmd/pm_enabled=true
# svcadm refresh ldmd
# svcadm restart ldmd
```
- On a Fujitsu SPARC system:
  - To disable PM manually, run the following commands on the target node:
```
# svccfg -s ldmd setprop ldmd/fjpm_enabled=false
# svcadm refresh ldmd
# svcadm restart ldmd
```
  - To enable PM manually, run the following commands on the target node:
```
# svccfg -s ldmd setprop ldmd/fjpm_enabled=true
# svcadm refresh ldmd
# svcadm restart ldmd
```
On Fujitsu SPARC systems, set the power-saving operation to disabled or off with the XSCF setpparmode command before ldm migrate-domain is executed. Subsequently, set the power-saving operation to elastic or on again after the migrated domain state is active.
- On a Fujitsu SPARC M12 system:
```
XSCF> setpparmode -p <ppar-id> -y -m powermgmt_policy=disabled
XSCF> setpparmode -p <ppar-id> -y -m powermgmt_policy=elastic
```
- On a Fujitsu M10 system:
```
XSCF> setpparmode -p <ppar-id> -y -m elastic=off
XSCF> setpparmode -p <ppar-id> -y -m elastic=on
```

Note - This issue primarily affects Fujitsu SPARC M12 and Fujitsu M10 systems, which have the Power Management feature enabled by default. Oracle systems do not have Power Management enabled by default, however it is possible to enable this feature on Oracle systems. If Power Management is enabled on an Oracle system, you can experience this issue.

PPAR DR(`deleteboard`) For a System Board Which Had a VF Failed

Bug ID 29363569: This issue may occur on Fujitsu SPARC M12 servers and Fujitsu M10 servers while executing PPAR DR deleteboard in the following scenario:

A domain has I/O devices.
A Virtual Function is configured on one or more of the I/O devices.
Execute a deleteboard command for the system board which contains the Virtual Function.

This issue occurs because a dependency on a future Oracle Solaris update was introduced into Oracle VM Server for SPARC 3.6.1. Until the Oracle Solaris update is available, use the following workaround.

Workaround: Remove I/O devices containing Virtual Functions manually prior to executing a deleteboard command.

The recommended operation for Fujitsu SPARC M12 servers and Fujitsu M10 servers is to remove I/O devices containing Virtual Functions manually prior to executing a deleteboard command.

Refer to the Fujitsu SPARC M12-2/M12-2S Service Manual or Fujitsu M10-4/Fujitsu M10-4S/SPARC M10-4/SPARC M10-4S Service Manual for details on PPAR DR.

Guest Ldom Bad Unexpected Error From Hypervisor Call at TL 1

Bug ID 27952673: As a component of the bug fix for 27952673, the migration-class1 definition no longer includes support for a 2-Gbyte page size because this page size is not available on SPARC M8 and SPARC T8 series servers. So, any migration that uses migration-class1 on a source machine that runs software prior to Oracle VM Server for SPARC 3.6 is blocked if the target machine is a SPARC M8 or SPARC T8 series server that runs at least the Oracle VM Server for SPARC 3.6 software.

To allow the migration, the domain on the source needs to inherit the new definition of migration-class1. To accomplish that, perform the following:

Upgrade to Oracle VM Server for SPARC 3.6 or later on the source system.
For any existing domains configured with cpu-arch=migration-class1, reboot the domain to inherit the new definition.

Oracle® VM Server for SPARC 3.6.1 Supplemental Release Notes