| Sun StorEdge 2 Gb FC PCI Dual Channel Network Adapter Installation Guide |     |
| Sun StorEdge 2 Gb FC PCI Dual Channel Network Adapter Installation Guide | |
| A P P E N D I X B |
|
Booting From the Network Adapter |
Booting from the network adapter requires special procedures in order to enable detection of installation devices. To boot from the small kernel image on the network adapter, you must use either a boot/install server or a temporarily directly-connected boot disk to enable bootability. The method using a boot install server is the Net Install Patching method and the method using a directly-connected boot disk is the Dump and Restore method. Both procedures should be performed by a knowledgeable UNIX system administrator.
The two methods are explained in this appendix in the sections:
The Dump and Restore method is more difficult than the Net Install Patching method and you might experience complications in the loss of the boot. However, the method is available to anyone with existing systems, whether small or large configurations, and it provides a disk image when you have completed the process.
The Net Install Patching method requires a jump-start server and network connectivity, and it is more suitable for large system configurations than for small ones. This method is easier than the Dump and Restore method and mitigates the risk of data loss.
Booting through a network adapter is mandatory when the host is an enterprise-level system or other system that does not have its own directly-connected disks. It is optional and preferable when a host has a directly-connected disk, you might prefer to boot through the adapter because the 2 Gb connection is faster.
This section contains a bootability method that is different from the one starting at "Invalid Cross-Reference Format". The topics in this section include:
The procedures in this section explain how to install the Solaris operating environment from a boot/install server onto the host with the host adapter. Any other host on the same subnet can be set up as a boot/install server.
|
Note - The procedure is identical whether the you are using a boot server or an install server.[1] Therefore, the convention used in this chapter is to refer to either type of server as the "boot/install server." |
Installing a client from a boot/install server uses two images of the Solaris operating environment:
This procedure assumes you know how to install Solaris software over the network as described in the Solaris installation manuals. For more information, see the man(1) pages for the commands that are used in this procedure.
You need to enable bootability using a boot/install server to make both images aware of the host adapter. At the start of the procedure, the system administrator downloads the driver and all required patches to an exported directory on the boot/install server.
The following steps explain the process. For actual instructions, begin with "Invalid Cross-Reference Format".
1. The boot image is copied from a Solaris CD-ROM or from another location onto a disk that is attached to a boot/install server.
2. The driver package is added to the boot image.
You add the driver package to the boot image so that the boot/install server can then send and receive data through the host adapter on the boot client.
3. The client boots from the boot/install server, the interactive suninstall(1M) application starts, and the system administrator provides configuration information requested at the prompts.
4. After the system administrator provides all configuration information requested by the installation program, the installation of the Solaris operating environment begins.
5. The install image is copied to the client.
6. Before a reboot, while the client is still booted from the boot image mini-root, the driver and any needed patches are copied from the boot/install server and then installed in the install image.
You install the driver package and patch the install image so that the host will be able to see the host adapter after a reboot.
|
Note - You can not use Jump Start until you modify the install image. |
7. The host boots from the boot disk through the network adapter.
1. Switch users to root on the host to be used as the boot/install server.
2. Use the setup_install_server(1M) command from the Tools directory in the location where the Solaris software resides.
As shown in the following screen example, the setup_install_server command copies the boot image to a directory on the boot/install server. (The boot directory is named /boot_dir /<original_OS_dir>/Boot in the example.) The example shows the command being run from the Tools directory on a mounted Solaris 8 installation CD-ROM.
# cd /cdrom/cdrom0/s0/Solaris_8/Tools # ./setup_install_server -t /<original_OS_dir>/Boot /<new_OS_copy_dir> |
3. Download the driver packages and the accompanying README file from Sun's download center into the /<export_public> directory on the boot/install server.
a. Go to the download center URL specified in the instructions on how to download the network adapter driver in the product notes.
c. Remove any previously-installed packages for this network adapter as instructed in the README.
d. Download the packages as described in the instructions in the release notes.
e. Use the uncompress(1M) and tar(1M) commands to uncompress and expand the packages in the tar file, as instructed in the README.
|
Caution - Do not use the pkgadd(1M) command line that is given in the READMEto install the packages. See "Invalid Cross-Reference Format"to install relocatable versions of the packages. |
4. Download the required patch(es) and the accompanying README file(s) from sunsolve.sun.com into the /<export_public> directory on the boot/install server.
a. See the release notes for a list of required patches.
b. Go to the URL specified in the instructions on how to download the network adapter driver required patch(es) in the release notes.
d. Remove any previously-installed related-patches that may be specified in the README.
e. Download the patch(es) as described in the instructions in the release notes.
f. Use the uncompress(1M) and tar(1M) commands to uncompress and expand the patch(es) in the tar file, as instructed in the README.
|
Caution - Do not use the patchadd(1M) command line that is given in the READMEto install the patch(es). See "Invalid Cross-Reference Format"to install relocatable versions of the patch(es). |
1. Install the driver packages into the boot image.
The following example installs all packages previously downloaded into the /<export_public> directory.
|
Note - Refer to the install_order file for the proper order in which to install the packages. You must follow this order for a successful installation so the driver will run. |
2. Install any needed patches into the boot image.
The following example installs any patch(es) that were previously downloaded into the /<export_public> directory. Repeat the patchadd command for all the patches you need to add.
3. Make sure the host name, its IP address, and its Ethernet address have been added to the name service (/etc files, NIS, or NIS+).
4. Run the add_install_client(1M) command to add the host with the host adapter as a boot/install client.
The example shows the add_install_client command followed by the name of the host followed by its platform name.
|
Note - You can find the platform name by running the uname command with the -m option on the host that has the host adapter. |
5. Log out of the boot/install server.
1. Bring the client host (with the host adapter) down to the ok prompt at run level 0.
See the Solaris system administration documentation for the commands that can be used with different configurations. The following example uses the shutdown(1M) command.
|
2. Boot the host from the net.
The Solaris interactive installation program runs from the boot/install server.
3. Respond to the prompts according to your configuration, as instructed in the Solaris installation guide.
Make sure to specify the new boot disk as the destination for the operating environment installation.
4. When prompted to choose between automatic reboot or manual reboot, click the Manual Reboot button, complete the remaining question, and start the installation.
This question offering a choice between automatic and manual reboot is the last question before the installation starts. If you are using the suninstall program, choose boot manual.
5. Mount the /<export_public> directory which contains the driver packages and any needed patches onto the /mnt directory mount point.
Enter the mount command followed by the hostname of the boot/install server, followed by a colon (:), followed by /<export_public> followed by /mnt. The following example uses boot_install_server as the name of the boot/install server.
6. Install the driver packages into the install image.
The following example installs all packages previously downloaded into the
/<export_public> directory. When prompted, install the packages in the order of SUNWsan, SUNWcfpl, SUNWcfplx.
7. Install any needed patches into the boot image.
The following example installs all patch(es) that were previously downloaded into the /<export_public> directory.
|
Note - Install the patches in their sequential numeric order to ensure the installation is successful. |
8. Bring the system down to the ok prompt at run level 0.
9. Reboot the host from the newly installed operating environment.
This section contains a bootability method that is different from the one starting at "Invalid Cross-Reference Format". The topics in this section include:
To enable bootability using a temporary boot disk, a boot disk must be directly connected, at least temporarily, to the host. The boot disk must have the following installed:
See the release notes for how to download and install the driver packages and any needed patches.
|
Note - The initial boot disk can be removed if it is not needed after the boot disk is enabled. |
|
Note - The examples in this section show disk 0 as the directly-connected boot disk, and disk 2 as the designated new boot disk that is connected through the host adapter. |
There are several sub procedures required to complete the first phase of enabling bootability from a temporarily connected boot disk. These sub procedures include:
1. Switch users to root on the host with the host adapter.
2. If the driver and any needed patches are not already installed, download the driver package from Sun's download center and install it on the host, following the instructions in the README file that comes with the driver.
To download the driver, follow the instructions in the release notes.
3. Reboot using the reboot(1M) command with the -r option.
After you log back into the host, you can record the layout of the partitions, or slices, on the system boot disk.
1. Enter the format(1M) command.
If needed, see the format man page and the instructions on adding a disk and using the format command in the Solaris administration documentation.
|
Note - These examples use disk 0 as the temporary disk (c0t0d0) and disk 2 (c7t16d0) as the new boot disk. |
2. Make a note of the device path name of the new boot disk.
For example, for disk 2, the new boot disk in this example, the device path name shown is:/pci@8,600000/SUNW,qlc@1/fp@0,0. You use this information later in Step 4 in "Invalid Cross-Reference Format".
|
Note - The qlc number changes, depending on the type of host and slot in use. |
3. Specify the disk where the operating system is installed on the temporary boot disk.
The following screen example specifies disk 2.
4. Enter the partition command to bring up the PARTITION MENU.
5. Enter the print command to display the partition table for the specified disk.
As shown in the example, the temporary boot disk has three slices defined: 0 (root), 1 (swap), and 7 (home) with sizes, 2.00GB, 1.00GB, and 1.00GB.
6. Record the layout (sizes and numbers) assigned to the slices on the temporary boot disk, and enter quit when done.
As shown in the previous example, the quit command returns you to the FORMAT MENU.
After you record the partition layout, change to the new boot disk.
1. At the format> prompt, type disk to change the current disk to the new boot disk.
2. Enter the number of the disk to be formatted after the disk command.
The following screen example uses disk 2. The FORMAT MENU appears.
3. Make a note of the device name of the disk.
The device name of the disk in the previous screen example is c7t16d0.
After you change to the new boot disk, specify one slice on the new boot disk for every slice on the temporary boot disk. The following examples specify the root slice 0 on the new boot disk to match slice 0 on the temporary boot disk.
1. Enter the partition command to bring up the PARTITION MENU.
2. Enter the number of the slice to be defined.
Slice 0 is specified in the following example. As shown, the partition table for the new boot disk displays.
3. Enter the partition ID tag.
The following example shows a question mark (?) entered after the prompt. The list of accepted partition id tags displays. The example then shows the default partition id tag of root accepted by pressing the Return key.
Enter partition id tag[root]: ? Expecting one of the following: (abbreviations ok): unassigned boot root swap usr backup stand var Enter partition id tag[root]: Enter partition permission flags[wm]: |
4. Enter the partition permission flags.
The following example shows the default permission flags wm accepted by pressing the Return key.
5. Enter the new starting cylinder.
The following example shows the default new starting cylinder of 0 accepted by pressing the Return key.
The following example shows the partition size of 2.00gb entered.
7. Enter the print command to display the updated partition table.
The following example shows that the root tag, the wm permissions flag, and the partition size of 2.00GB are assigned to slice 0.
8. Repeat Step 2 through Step 7 as needed until all slices are defined as they are in the temporary boot disk.
9. Enter the quit command to return to the FORMAT MENU.
After you specify the slices on the new boot disk, label the new boot disk with the new partition table.
3. When the labeling is complete, enter q[uit] to quit the format program.
Create a file system on each slice on the disk using the newfs(1M) command.
Enter the newfs command followed by the device name of the slice. In this example, the device name for slice 0 of disk c7t16d0 is /dev/rdsk/c7t16d0s0.
For more information, see the section on how to create file systems in the Solaris system administration documentation.
Do this step to create a file system on the new boot disk for every slice on the temporary boot disk. When you are finished, go to "Invalid Cross-Reference Format".
This section has several sub procedures, including:
1. Install the boot block on the root (/) file system of the new disk.
The following example uses the installboot(1M) command to install the boot block. The boot block resides in the /usr/platform/platform_name/lib/fs/ufs/bootblk directory. The example shows invoking the uname command with the -i option between left single quotes on the command line to specify the platform name.
For more information, see the instructions on how to install a boot block in the Solaris system administration documentation.
2. Mount the root file system from slice 0 of the new boot disk onto the /mnt mount point.
3. Use the ufsdump(1M) and ufsrestore(1M) commands to copy the contents of the root file system from the temporary boot disk to the root slice of the new boot disk (on the /mnt mount point).
4. Unmount the root file system on slice 0 from the /mnt mount point.
After you copy the boot block and root files, update the vfstab file.
1. Mount the root file system from slice 0 of the new boot disk onto the /mnt mount point.
2. Change directories to /mnt/etc and open the vfstab(4) file for editing.
The following example shows the file systems defined.
# cd /mnt/etc # vi vfstab ... /dev/dsk/c0t0d0s1 - - swap - no - /dev/dsk/c0t0d0s0 /dev/rdsk/c0t0d0s0 / ufs 1 no - /dev/dsk/c0t0d0s7 /dev/rdsk/c0t0d0s7 /home ufs 2 yes - |
3. Replace the name of the temporary boot disk with the name of the new boot disk, and then save and quit the file.
The following example shows the disk name c0t0 changed to c3t8 in the mount table entries for slices 0, 1, and 7.
/dev/dsk/c7t16d0s1 - - swap - no - /dev/dsk/c7t16d0s0 /dev/rdsk/c7t16d0s0 / ufs 1 no - /dev/dsk/c7t16d0s7 /dev/rdsk/c7t16d0s7 /home ufs 2 yes - :wq # |
1. Mount the file system onto the /mnt mount point.
This example shows the copying of the /home file system from slice 7 to the new boot disk.
2. Use the ufsdump(1M) and ufsrestore(1M) commands to copy the contents of the file system from the temporary boot disk to the new boot disk.
3. Unmount the file system from the /mnt mount point.
4. Repeat Step 1 through Step 3 as needed until you have copied all the file systems' contents to the new boot disk. When finished, go to "Invalid Cross-Reference Format".
1. Bring the host with the host adapter down to the ok prompt at run level 0.
See the Solaris system administration documentation on shutting down a host for the commands that can be used with different configurations. The following screen example uses the shutdown(1M) command.
2. Use the nvalias command to alias the device name of the disk to a short name for the disk.
The following example uses /pci@8,600000/SUNW,qlc@1/fp@0,0/disk@10,0, which was the device path name for disk 2 in "Invalid Cross-Reference Format".
3. Use the nvstore command to store the new alias followed by the reset all command.
4. Define the new boot disk as the default boot-device.
Use the data gathered in Step 2 on "Invalid Cross-Reference Format".
a. Enter the setenv command followed by the boot-device parameter followed by the name of the new disk.
5. Enter the boot command with the -r option so that the Solaris operating environment can recognize the adapter.
| Sun StorEdge 2 Gb FC PCI Dual Channel Network Adapter Installation Guide | 816-5001-12 | |
Copyright © 2007, Sun Microsystems, Inc. All Rights Reserved.