System Boot, Recovery, and Platform Changes

The system boots from a ZFS root file system in Oracle Solaris 11. By default, the ZFS root file system is contained within a ZFS root pool, named rpool. Creating a UFS file system is still supported, but you cannot boot from a UFS or a Solaris Volume Manager root file system in this release.

Review the following information that impacts the way the system is booted for recovery purposes:

If you use a system's service processor (SP) or ILOM to recover from a system problem, accessing a system's SP or ILOM is identical to previous releases. The differences mostly pertain to how the system is booted after you get to a SPARC based system's ok PROM prompt or to an x86 based system's firmware screen (BIOS or UEFI).
In Oracle Solaris 10, you use flash archive features to create a copy of a UFS or ZFS root environment and then restore the flash archive to recover the system environment, in the case of a system or device failure.

In Oracle Solaris 11, the system recovery process includes the following steps:
- Archiving the root pool snapshots on a remote system
- Replacing any failed system component or device
- Recreating the root pool and setting the bootfs property
- Restoring the previously archive root pool snapshots
- Manually installing the boot blocks
See Chapter 11, Archiving Snapshots and Root Pool Recovery, in Oracle Solaris 11.1 Administration: ZFS File Systems.
When attempting to boot a system for recovery purposes, if the system cannot be booted, but the failure is not because the root pool is unavailable, you can use new boot options from the installation media or from an installation server to resolve the problem. See Booting for System Recovery.

GRUB, Firmware, and Disk Labeling Changes

Starting with Oracle Solaris 11.1, the following changes are introduced:

GRUB 2 is the default boot loader on x86 platforms – GRUB 2 replaces the original GRUB 0.97-based boot loader (GRUB Legacy). GRUB 2 fully supports booting from disks that are larger than 2TB. GRUB 2 also supports the Unified Extensible Firmware Interface (UEFI) and the GUID Partition Table (GPT) partitioning scheme.
GRUB menu changes – Unlike the editable menu.lst file that is used by GRUB Legacy, GRUB 2 uses a configuration file named grub.cfg that is syntactically different than the legacy menu.lst file. The grub.cfg file stores most of the GRUB configuration and is managed solely by using the bootadm command. To accommodate this change, the bootadm command has been expanded to include several new subcommands and a new -P option that enables you to administer the GRUB configuration for multiple root pools.

Note - Because GRUB configuration changes can automatically overwrite changes that are made to the grub.cfg file, do not manually edit this file. See Chapter 2, Administering the GRand Unified Bootloader (Tasks), in Booting and Shutting Down Oracle Solaris 11.1 Systems and bootadm(1M).
Management of non-Oracle Solaris boot entries – GRUB 2 includes an additional configuration file named custom.cfg. This file can be used to add custom menu entries to the GRUB configuration. The custom.cfg file does not exist on the system by default. You must create the file and it must be stored in the same location as the grub.cfg file (/pool-name/boot/grub/). During the boot process, GRUB checks for the custom.cfg file in the top-level dataset of the root pool (boot/grub). If the file exists, GRUB sources the file and processes any commands within the file as if the contents were actually part of the grub.cfg file. See Customizing the GRUB Configuration in Booting and Shutting Down Oracle Solaris 11.1 Systems.
64-Bit UEFI firmware support – Oracle Solaris now supports x86 based systems with 64-bit UEFI firmware. An installation on UEFI firmware is supported through the DVD, USB, and network installation methods. UEFI version 2.1+ is required.

If you are booting a system with UEFI firmware from the network, the boot process has changed slightly. See Booting Systems With UEFI and BIOS Firmware From the Network in Booting and Shutting Down Oracle Solaris 11.1 Systems for more details.
Booting from GPT labeled disks– GPT labeled disks are now supported on both SPARC and x86 platforms. Installing Oracle Solaris 11.1 on an x86 or SPARC based system with GPT-aware firmware applies a GPT disk label on the root pool disk that uses the entire disk in most cases. Otherwise, installing Oracle Solaris 11.1 on a SPARC based system applies an SMI (VTOC) label to the root pool disk with a single slice 0.

For SPARC based systems that support a GPT labeled boot disk, see Support for GPT Labeled Disk Added on SPARC Platforms in Oracle Solaris 11.1 Release Notes for more information about how to apply the GPT-aware firmware update.

If you are running a release that supports GRUB Legacy and are moving to a release that supports GRUB 2, see Upgrading Your GRUB Legacy System to a Release That Supports GRUB 2 in Booting and Shutting Down Oracle Solaris 11.1 Systems.

Booting for System Recovery

The following error and recovery scenarios are similar to previous releases:

You can use the boot -a command to bypass a problem in the /etc/system file. When prompted, use syntax that is similar to the following:
```
Name of system file [/etc/system]:  /dev/null
```
Press Return at the other prompts, as needed.
A backup BE is created automatically during most pkg update operations. This feature enables you to boot back to a previous BE in case some error occurs during the image update process. Consider creating a backup BE before you make a system configuration change.
```
# beadm create solaris-backup
# beadm list
BE             Active Mountpoint Space  Policy Created          
--             ------ ---------- -----  ------ -------          
solaris        R      -          4.01G  static 2013-02-08 16:53 
solaris-backup N      /          47.95M static 2013-02-11 10:48 
```
See How to Boot From a Backup BE for Recovery Purposes for the steps to boot from a backup BE.
Boot from the installation media or from an install server over the network to recover from a problem that is preventing the system from booting or to recover from a lost root password.

On SPARC based systems, the boot net:dhcp command replaces the boot net command that is used in Oracle Solaris 10 releases.
Boot a system in single-user mode to resolve a minor problem, such as correcting the root shell entry in the /etc/passwd file or changing a NIS server.
Resolving a boot configuration problem generally involves importing the root pool, mounting the BE, and fixing the problem, for example reinstalling a corrupt x86 boot loader.

How to Boot From a Backup BE for Recovery Purposes

Booting the failsafe archive is no longer supported on SPARC and x86 platforms. Whenever possible, use up-to-date backup BEs for recovery purposes. BEs are bootable instances of the Oracle Solaris image, plus any other application software packages that are installed into that image. Multiple BEs reduce risk when updating software because the backup BE preserves the original BE.

You can create a new BE based on an active or an inactive boot environment. Or, you can create a new BE based on a clone of your original BE. A clone copies the root dataset and everything hierarchically under the main root dataset of the original BE. See Creating and Administering Oracle Solaris 11.1 Boot Environments.

If the system does not boot from the active BE, select a backup BE from which to boot.

Boot from a backup BE, as follows:

SPARC: Boot the system so that you can select an alternate or backup BE.
1. Boot with the boot -L command.
```
ok boot -L
```
2. Select an alternate or backup BE.
```
Boot device: /pci@7c0/pci@0/pci@1/pci@0,2/LSILogic,sas@2/disk@0,0:a  
File and args: -L
1 Oracle Solaris 11.1 SPARC
2 solaris-backup
Select environment to boot: [ 1 - 2 ]: 2
```
  In the previous output, the active BE is Oracle Solaris 11.1 SPARC, which most likely does not match the actual BE name, but it represents the current BE.
3. Boot the backup BE.
  After you select the BE from which to boot, identify the on-screen boot path and type that information at the prompt.
```
To boot the selected entry, invoke:
boot [<root-device>] -Z rpool/ROOT/solaris-backup

Program terminated
{0} ok boot -Z rpool/ROOT/solaris-backup
```
  If the system does not boot, review the additional boot recovery steps in How to Boot the System For Recovery Purposes.

x86: Boot the system to identify the alternate or backup BE from the GRUB menu.

When the GRUB menu is displayed, identify the backup BE.

GNU GRUB  version 1.99,5.11.0.175.1.0.0.14.0

 ****************************************************************************
 *solaris                                                                   *
 *solaris-1                                                                 *
 *                                                                          *
 *                                                                          *
 *                                                                          *
 *                                                                          *
 ****************************************************************************

Select the backup BE, then press Return to boot that entry.
If the system does not boot from the backup BE, review the additional boot recovery steps in How to Boot the System For Recovery Purposes.

How to Boot the System For Recovery Purposes

Select the appropriate boot method.
Note - On x86 platforms, use the reboot command with the -p option to initiate a standard reboot of the system, which enables you to view the GRUB menu or select an installation option. Otherwise, the system defaults to a fast reboot.
- x86: Live Media – Boot from the installation media and use a GNOME terminal for the recovery procedure.
- SPARC: Text installation – Boot from the install media or from the network, and select option 3 Shell from the text installation screen.
- x86: Text installation – From the GRUB menu, select the Text Installer and command line boot entry, then select the option 3 Shell from the text installation screen.
- SPARC: Automated installation – Use the following command to boot directly from an installation menu that allows you to exit to a shell.
```
ok boot net:dhcp
```
- x86: Automated installation – Boot from an install server on the network that supports a PXE boot. Select the Text Installer and command line entry from the GRUB menu. Then, select the option 3 Shell from the text installation screen.
For example, after the system is booted, select option 3 Shell:
```
        1  Install Oracle Solaris
        2  Install Additional Drivers
        3  Shell
        4  Terminal type (currently xterm)
        5  Reboot

Please enter a number [1]: 3
To return to the main menu, exit the shell
#
```

Select from the following boot recovery problems:

Resolve a bad root shell by booting the system to single-user mode and correcting the shell entry in the /etc/passwd file.

On x86 based systems, boot the system in single-user mode by editing the selected boot entry in the GRUB menu. Add the -s option to end of the $multiboot line.
```
$multiboot /ROOT/s11u1_24b/@/$kern $kern -B $zfs_bootfs -s
```

On SPARC based systems, shut down the system and boot in single-user mode. After you log in as root, edit the /etc/passwd file, and fix the root shell entry.

# zpool import -f rpool
# beadm list
be_find_current_be: failed to find current BE name
BE        Active Mountpoint Space Policy Created          
--        ------ ---------- ----- ------ -------          
solaris   -      -          7.74M static 2013-02-09 09:40 
solaris-1 R      -          4.08G static 2013-02-13 07:24 
# mkdir /a
# beadm mount solaris-1 /a
# TERM=vt100
# export TERM
# cd /a/etc
# vi shadow 
<Carefully remove the unknown password>
# cd /
# beadm umount solaris-1
# halt

# init 0
ok boot -s

Boot device: /pci@780/pci@0/pci@9/scsi@0/disk@0,0:a File and args: -s
SunOS Release 5.11 Version 11.1 64-bit
Copyright (c) 1983, 2012, Oracle and/or its affiliates. All rights
reserved.
Booting to milestone "milestone/single-user:default".
Hostname: tardis.central
Requesting System Maintenance Mode
SINGLE USER MODE

Enter user name for system maintenance (control-d to bypass): root
Enter root password (control-d to bypass): xxxxxxx
single-user privilege assigned to root on /dev/console.
Entering System Maintenance Mode

Feb 13 00:21:31 su: 'su root' succeeded for root on /dev/console
Oracle Corporation SunOS 5.11 11.1 September 2012
su: No shell /usr/bin/mybash.  Trying fallback shell /sbin/sh.
root@tardis.central:~# TERM=vt100; export TERM
root@tardis.central:~# vi /etc/passwd
root@tardis.central:~# <Press control-d> 
logout
svc.startd: Returning to milestone all.

x86: Resolve a corrupt boot loader problem by reinstalling it, as follows:
1. Follow the directions in Step 1 to boot the system from media or from the network, then import the root pool.
```
# zpool import -f rpool
```
2. Reinstall the boot loader.
```
# bootadm install-bootloader -f  -P pool-name
```
  where -f forces the installation of the boot loader and bypasses any versioning checks for not downgrading the version of the boot loader on the system. The -p option is used to specify the root pool.
  
  Note - Do not used the -f option unless you are sure that you want to overwrite the boot loader with the version that is on the media. See Installing GRUB 2 by Using the bootadm install-bootloader Command in Booting and Shutting Down Oracle Solaris 11.1 Systems.
3. Export the root pool.
```
# zpool export pool-name
```
4. Reboot the system.

Resolve an unknown root password problem that prevents you from logging into the system.

Follow the instructions in Step 1 to boot from media or from the network, then import the root pool (rpool) and mount the BE to remove the root password entry.
This process is identical on SPARC and x86 platforms.

Set the root password by booting in single-user mode and setting the password.

This step assumes that you have removed an unknown root password in the previous step.

On x86 based systems, edit the selected boot entry in the GRUB menu, adding the -s option at the end of the $multiboot line.
```
$multiboot /ROOT/s11u1_24b/@/$kern $kern -B $zfs_bootfs -s
```

On SPARC based systems, boot the system to single-user mode, log in as root, and set the root password. For example:

ok boot -s

Boot device: /pci@780/pci@0/pci@9/scsi@0/disk@0,0:a File and args: -s
SunOS Release 5.11 Version 11.1 64-bit
Copyright (c) 1983, 2012, Oracle and/or its affiliates. All rights reserved.
Booting to milestone "milestone/single-user:default".
Hostname: tardis.central
Requesting System Maintenance Mode
SINGLE USER MODE

Enter user name for system maintenance (control-d to bypass): root
Enter root password (control-d to bypass): <Press return>
single-user privilege assigned to root on /dev/console.
Entering System Maintenance Mode

Feb 13 00:58:42 su: 'su root' succeeded for root on /dev/console
Oracle Corporation      SunOS 5.11      11.1    September 2012
root@tardis.central:~# passwd -r files root
New Password: xxxxxx
Re-enter new Password: xxxxxx
passwd: password successfully changed for root
root@tardis.central:~# <Press control-d> 
logout
svc.startd: Returning to milestone all.

Boot, Platform, and Hardware Changes

Note the following boot, platform, and hardware feature changes in Oracle Solaris 11:

x86 platform support is 64–bit only – Support for booting a 32–bit kernel on x86 platforms has been removed. Systems that have 32-bit hardware must either be upgraded to 64–bit hardware or continue to run Oracle Solaris 10. Note that 32–bit applications are not impacted by this change.
Bitmapped console support – Oracle Solaris 11 includes support for high resolution and color depth consoles. By default, your machine will boot with a 1024x768x16-bit console, unless your video card does not support this setting. In which case, the setting will fall back to 800x600, then finally to 640x480. The console type (and also the older VGA TEXT 640x480 console) can be controlled through both kernel parameters and through options that you specify by editing the GRUB menu at boot time, as follows:
```
-B console={text|graphics|force-text}
```
See Redirecting the Oracle Solaris Console at Boot Time in Booting and Shutting Down Oracle Solaris 11.1 Systems.
Fast Reboot support on x86 and SPARC platforms – On x86 platforms, Fast Reboot implements an in-kernel boot loader that loads the kernel into memory and then switches to that kernel. For SPARC based systems that support the Fast Reboot feature, the boot process is accelerated by skipping certain POST tests.

The Fast Reboot feature works differently on SPARC platforms than it does on x86 platforms. To initiate a fast reboot of a SPARC based system, use the -f option with the reboot command. Because Fast Reboot is the default behavior on x86 platforms, the -f option is not required. Use either the reboot command or the init 6 command to initiate a fast reboot of an x86 based system. The Fast Reboot feature is managed through SMF properties that can be enabled or disabled, as required. For more details, see Accelerating the Reboot Process in Oracle Solaris Administration: Common Tasks.
Removal of support for the SPARC sun4u architecture – With the exception of the M-series (OPL) hardware, you cannot boot Oracle Solaris 11 on the sun4u architecture. If you attempt to boot Oracle Solaris 11 on one of these systems, the following error message is displayed:
```
Rebooting with command: boot
Error: 'cpu:SUNW,UltraSPARC-IV+' is not supported by this release of Solaris.
NOTICE: f_client_exit: Program terminated!
```

Skip Navigation Links
Exit Print View
	Transitioning From Oracle Solaris 10 to Oracle Solaris 11.1 Oracle Solaris 11.1 Information Library