6.1 OVERVIEW AND CAPABILITIES  Back To Top

The System Crash AIR-BAG(TM) allows you to restore your entire filesystem exactly as it was previously set up. All you need for this are the Air-Bag diskettes, the Master backup, and your last Incremental Tape Backup.

You will find the Air-Bag diskettes to be of great reassurance. The knowledge that your data is secure even with a total system crash; and that even in the worst case, it can simply be transferred to another machine, is true security.

The Air-Bag diskettes can be used as utility diskettes by advanced users to salvage data from your hard disk in the event of a crash.

With this system you can do the following:

• Restore your entire filesystem to the same UNIX partition
• Enlarge your UNIX partition and restore your SCO UNIX filesystems to the original format
• Remap out BAD TRACKS that may have gone bad since the original installation, then restore your filesystem.
• Enlarge the DOS partition (if you are using one) and move the SCO UNIX partition further out on the hard disk.
• Transfer your entire filesystems to a new hard disk (useful in case of hard disk crash)
• Transfer your entire filesystems to a new machine (useful to install on a loaner machine while yours is being repaired or to distinguish hardware versus software bugs and the cause of erratic behavior)
• View the Bad Tracks that were mapped out on your hard disk even if the hard disk has crashed.
• Restore your root filesystem only (useful when it has become corrupted).
• Make your hard drive bootable
• Restore your previous CMOS settings (if they are lost during a power failure)
• Run a diagnostics/system confidence test
• Dial in through a modem and perform any of the above tasks

6.2 INSTALLING THE SYSTEM CRASH AIR-BAG�  Back To Top

The System Crash AIR-BAG is installed by extracting all the files from the installation diskette using xinstall (UNIX and OpenServer 5) or install (XENIX only). Most often this occurs automatically with the installation of the backup product that comes with the Air-Bag. If you are installing an updated version, be careful that the Air-Bag is fully installed before generating your A1 and A2 diskettes. It is possible to have only installed the backup software portion of the upgrade, and invoke the Air-Bag from within the backup software menu system. However, if you are not careful, you may have invoked the old version since the upgraded files have not yet been installed. Therefore, if upgrading, be sure all products have been fully installed before invoking the Air-Bag diskette generator.

During installation your hard disk is checked for the presence of the 1024 Cylinder Barrier problem. If you are not running SCO OpenServer 5 (OS5), you can choose to save 250 blocks of space on your A2 diskette.

6.2.1 Creating Extra Space on the A2 Floppy  Back To Top

For UNIX 3.2v4 you can save 250 blocks of space on the A2 floppy by answering a few questions during installation and inserting the 2 floppy. The entire operation takes about 15 seconds and never needs to be done again. From that point on the A2 floppy has 250 more blocks of free space. Do this if you want to put additional utilities on the A2 floppy.

6.2.2 Adding XENIX Filesystem Support to OS5  Back To Top

During the first installation when running under OpenServer 5, you are presented with the option of adding XENIX filesystem support to the kernel. By default, OpenServer 5 does not have the XENIX filesystem support enabled. It is preferable that you add this support for the following two reasons:

1) The A1 and A2 diskettes will boot three times faster (2 min vs. 6 min).

2) The totally automated restore features works better with this capability in the kernel. It can then reliably reproduce any combination of filesystems. Most of the testing of the totally automated restore has been with XENIX filesystem support enabled.

If you choose to add this support (which is strongly recommended), changes are made to the link kit to enable this support. Then you opt to relink the kernel immediately or postpone it to a later time. Adding the XENIX filesystem support adds only 47K to the size of the kernel.

6.3 CREATING THE AIR-BAG DISKETTES  Back To Top

6.3.1 Creation Script  Back To Top

The Air-Bag diskettes are created by using the script /etc/airbag. This script is loaded into the /etc directory during installation of the package. This script creates your A1 and A2 diskettes, from which you boot in an emergency to restore your system. You should be in multi-user mode when you create the A1 and A2 diskettes so that the system configuration under normal operating circumstances can be captured and then later restored when needed. The main creation menu appears as follows:

FIGURE 12. AIR-BAG Creation Menu

Each step of the creation of the Air-Bag diskettes is outlined in detail on the console screen. Most steps report a success or failure after completion of the step. At the end of all steps, this can be reviewed to check for error messages. Remember two important things when creating your diskettes. First, you should always make a spare copy of your A2 diskette. Second, you should always boot from the newly made A1 and A2 diskettes and test the tape and hard disk accessibility.

6.3.2 Choice of Diskettes  Back To Top

• The /etc/airbag creation script lets you choose the media type for the boot diskettes. You can choose any of the following
• 1.44 Mb 3.5 inch diskette
• 2.88 Mb 3.5 inch diskette
• 1.2 Mb 5.25 inch diskette

6.3.3 Customizing the Questions Asked

The creation of each diskette is designed so that all the questions are asked up front, then the Air-Bag diskette generator proceeds to create the diskette according to the answers to the questions you have supplied. However, in order to save you time and ensure consistency, many of the questions can be adjusted in one of four ways.

 

Question never appears

Question always appears

Question appears and the default answer is y (yes)

Question appears and the default answer is n (no)

 

The questions that are customizable in this way are represented by an environment variable that is present at the top of the script /etc/airbag. The environment variable can be set to one of four values: {YES, NO, ALWAYS, NEVER}. For example, the environment variable FLOPPY_TAPE represents the question that asks if you have a floppy based tape controller. For the majority of installations the answer to this questions is no. If you do not want to see this question at all (since it can be confusing for the individual creating the diskettes), set the environment variable in the script file /etc/airbag as follows:

: {FLOPPY_TAPE:=NEVER}

This ensures that you will never see this question again. If you do want to see the question and want the default answer to be y (yes) set it as follows:

: {FLOPPY_TAPE:=YES}

See the section entitled Adjusting Custom Environment Variables on page 138 for details on the types of questions that are available for customizing.

6.4 CREATING THE A1 DISKETTE (BOOT DISK)  Back To Top

The A1 diskette contains the kernel and floppy based boot straps. When this diskette is created, you can add support for Boot Time Loadable Drivers (BTLD) or not. Additionally, you select the kernel you want to use. The following sections explain this in more detail.

6.4.1 Boot Time Loadable Drivers  Back To Top

A Boot Time Loadable Driver (BTLD) is a driver that is loaded dynamically from a special diskette at boot time. This BTLD driver is provided by the manufacturer of a special hard disk drive or controller card. For example, if you have a special third party RAID controller, it may come with a BTLD driver. BTLD drivers are necessary since SCO cannot anticipate in advance all the different types of hardware that will be developed after a release of the SCO UNIX Operating System. Instead, SCO provides the hardware manufacturer with the ability to link the BTLD driver for their hardware into the operating kernel at boot time. If the piece of hardware is critical for booting (accessing main hard drive) or restoring data (tape drive) it must either be supported natively by the operating system or come with a BTLD driver.

Support for linking in BTLD drivers at boot time is available by answering y to the following question:

You can now enable the ability to load BTLD diskettes.

BTLD diskettes are Boot Time Loadable Diskettes and allow you to boot with hard drives not otherwise recognized by the standard SCO UNIX.

Enabling this feature requires considerable space on the A1 diskette and may cause you to run out of space.

If you do not understand this question, answer `n'

[Default: no] Do you wish to enable the ability to load BTLD diskettes? (y/n):.

Enabling BTLD support takes more space on the A1 diskette. It is possible that if you have a large kernel, you may not be able to fit both BTLD support and the large kernel. If this is the case, you have to make a decision as to which is more important or simply make two A1 diskettes, each with slightly different capabilities. In this document, the term BTLD enabled A1 refers to an A1 diskette that has been generated with BTLD support.

If you are booting from a BTLD enabled A1 diskette, you can activate the BTLD feature by typing the word link from the boot: prompt. After the computer reads the A1 diskette for a while, it will ask you to insert your BTLD driver diskette. The BTLD driver is then linked into the booted kernel. This allows you to access devices otherwise not accessible.

Once you restore from a tape, if the BTLD driver is not linked into the kernel on the newly restored system, you can boot from the hard disk (OpenServer 5) by typing:

hd(40)unix root=hd(42)swap=hd(41)link=xxxx btld=fd(64) Sdsk=auto(0,0,0)

where xxxx is the name of the BTLD driver. For example, the SCSI driver for NCR 710 chip is called bscam. In this case, the word xxxx is replaced by the word bscam. The keyword auto is for safety so that the kernel knows to look for any possible SCSI disk drivers. Keep in mind, the name of the BTLD disk is different from the name of the actual driver. Sometimes this is confusing. For example, the bscam BTLD disk has a driver called bhba. You could accidentally use the term bhba instead of bscam.

In the above example, you can alternatively put Sdsk=bhba(0,0,0) on the command line instead of auto(0,0,0). If the controller is a secondary controller, the line should read

Sdsk=bhba(1,0,0).

However, now that this is all said and done, it is preferable to load the BTLD driver on a working system in advance using installpkg and to link it into the kernel. When you generate your A1, use this newly built kernel. Then make your Master backup tape. If you can, boot the A1 on the computer for which the BTLD driver is intended and perform the verification stage (OPTION 9 verifies tape and hard drive), then after you restore from a crash, the restored kernel should recognize the new hard disk.

6.4.1.1 Trouble-Shooting BTLD Problems  Back To Top

If you see the following message after booting from the A1 and A2 diskettes and inserting your BTLD driver diskette:

xxxx: Can't find required symbol "nulldev"

and a host of messages similar with "nulldev" replaced by other names, this means you forgot to enable BTLD support when you generated the A1 diskette. You have to regenerate an A1 disk with BTLD support enabled. If you are not sure whether or not BLTD is enabled, you can mount the A1 disk and look for the file /BTLD_enabled in the root directory.

If you link in a BTLD driver before loading the driver you see the message:

Bad decode table

This means that the loading of the kernel was corrupted. Most often this is due to a key being hit, or the CPU speed being toggled by a turbo button while the kernel is loading. Try booting again and it should work.

6.4.2 Selecting the Boot Kernel  Back To Top

The Air-Bag diskette generator allows a high degree of flexibility in choosing the UNIX kernel from which to boot. This is extremely useful in large environments where the UNIX kernel is too large to fit on one floppy diskette.

You are given several choices regarding the UNIX kernel to place on the A1 diskette (boot diskette). The menu appears as follows:

Now to choose the UNIX kernel to put on BOOT Diskette...

Currently: ("/unix")

Generally speaking, you should select one of the choices that is
preceded by an arrow "**>"

Please choose:
**> 1) Put on basic UNIX N1 Kernel ("/unix.N1")
**> 2) Put on current configured UNIX kernel ("/unix")
3) Put on customized UNIX kernel ("/unix.airbag")
4) Type in a specific UNIX kernel name from keyboard
5) See more information about above choices
99) Exit
Enter:

You should always try to use the most currently configured UNIX kernel first. If this is too large, then try the /unix.N1 kernel. If this doesn't work because of third party hard disk or tape drivers that your hardware requires, then you must make a customized version of the UNIX kernel. Please refer to the next section in this document that specifically refers to customizing your boot kernel.

OPTION 4 allows you to enter from the keyboard the name of a UNIX kernel to use. This is useful if you are trying different customized kernels and want to keep them separate. OPTION 5 contains over six help screens of information that will be useful in making a decision.

6.4.3 Customizing your Boot Kernel  Back To Top

Air-Bag allows you to use a customized kernel, /unix.airbag, for special needs. Most commonly, this is used to add support for multiple SCSI hard disk controller cards. See Adding SCSI Controller Support to your Kernel on page 162 for details.

Another reason to use a customized kernel is to make a smaller one. This is not needed for SCO OpenServer 5 (Air-Bag boots from a compressed kernel); however, for SCO UNIX 3.2v2 and 3.2v4 it may be needed if you have a lot of drivers in the kernel. See What to do if your Kernel is too Large on page 168 for details.

Once you decide on the method of customizing and creating a custom kernel using the LINK kit, the kernel can be copied to the file /unix.airbag. If this has been done, Air-Bag recognizes the presence of this kernel and recommends that it be used during bootable diskette creation.

You can test the kernel first before generating an Air-Bag diskette by booting from the hard disk and typing the following at the boot prompt:

SCO System V

Boot: hd(40)unix.airbag

If this boots, and you can access your hard disk and tape drive, then it can be used to create the A1 diskette (boot diskette). Again, as always, you should boot from Air-Bag diskettes and test the tape drive.

You can now tell the name of the kernel that is on the A1 floppy. Previously, all kernels (unix.N1, unix.airbag, unix) were called unix. Now there is an additional link to the originating name. This is handy if you don't know with which type of kernel the diskette was made

6.5 CREATING THE A2 DISKETTE (ROOT FILESYSTEM)  Back To Top

The A2 diskette contains all the necessary files to run the Air-Bag. This includes the system utilities as well as the Airbag Menu System. Since the A2 diskette creation is so flexible, there are several questions that must be answered up front. The creation of the A2 diskette was designed so that all questions are asked first, then the Air-Bag diskette generator creates the A2 diskette accordingly.

6.5.1 Selecting Hard Disk Controller Type  Back To Top

When the A2 diskette is generated, you have an opportunity to indicate any special equipment that you may have. This includes an OMTI 8260/8270 controller or a DPT SCSI controller. The Air-Bag diskette generator attempts to determine the type of equipment that you have. It may not be correct in all circumstances so you should verify it when you see the following prompt:

Please select the controller installed (MAIN HARD DISK):
1. SCSI controller
2. DPT SCSI Series 2011/2012 controller
3. OMTI 8260 /8627 controller
4. IDE controller - Error free
5. IDE controller - with Bad tracks (e.g. Seagate.)
6. ST-506 or compatible
7. Compaq IDA controller
8. None of the above/Not sure
Select:

If you have an ESDI hard disk using either an Adaptec controller or a Western Digital controller, choose OPTION 8. These controllers appear as an ST-506 controller to the computer.

The answer to this question determines how Air-Bag prepares your hard disk before restoring the filesystems. It is important to answer this question accurately. This information can be changed when you are booted from Air-Bag diskettes, but it is not recommended that you do so unless you have just changed the hard disk controller type. For example, if your computer has a hard disk crash, and you install a hard disk controller of a different type than was previously installed, you can then change this to the new type from the main menu when booted from the Air-Bag diskettes.

Even though the IDE standard states that the disk drive should behave as if it is free of bad sectors, there are some manufacturers who have failed to adhere to this standard. If your IDE drives comes with a hard disk flaw map, then your IDE drive is one of these. In this case you should select OPTION 5 on the above sample menu. This allows you to scan the disk for bad sectors/tracks when it is initialized. Many Seagate IDE drives are of this type. The Conner IDE drives are all error free.

6.5.2 Floppy Controller Based Tape Drives  Back To Top

Support has been included for tape drives that are connected to a floppy disk controller. These are most often the Colorado Memory Jumbo series and the QIC-40 or QIC-80 Mini-cassette tape drives.

If you have one of these tape drives, you must answer y to the question regarding floppy controller based tape drives. This determines how Air-Bag restores data to your hard disk. Since these tape drives cannot work simultaneously with a floppy disk, you cannot restore the filesystems when booted from the Air-Bag diskettes. Air-Bag must be informed of this fact, so that it can take special action.

Under OpenServer 5, floppy controller based tape drives run without any special support since the Air-Bag Menu System runs from a memory based RAM filesystem rather than the floppy itself. Therefore, this question is not presented when you are running under OpenServer 5.

If you have an Irwin mini-cassette tape, you should answer n, since support for this is already built in to the Air-Bag. If the device file /dev/mcdaemon is found, it is assumed you are using an Irwin mini-cassette tape drive. The daemon program /etc/mcdaemon is invoked automatically when you are booted from the A1 and A2 diskettes.

If you have a separate controller card for the tape drive, then you answer n to the question regarding floppy based tape drive controllers. For example, the Colorado Memory Systems Jumbo tape drive has an optional controller that you can purchase. This allows you to use two floppy disk drives. In this case, for the purposes of the Air-Bag, you do not have a floppy based tape controller and you should answer n to the question regarding such.

If you are not sure whether or not your tape drive uses the floppy controller, you can answer n to the question and generate the Air-Bag diskettes. Then boot from this and test out the tape unit. If it does not work properly then you can regenerate the Air-Bag diskettes by answering y to the question.

6.5.3 NFS Mounted Filesystems  Back To Top

The Air-Bag diskette generator attempts to exclude all NFS mounted filesystems from the list of mounted filesystems that it generates. However, you should double check this list yourself. The list is presented to you on the screen. The list is put in the file /etc/AIRBAG.MOUNTS.

6.5.4 Special Device Files  Back To Top

The creation script /etc/airbag, copies almost everything in the /dev directory to the A2 diskette, so all your device files (and thus drivers) are available when you boot from the floppy. It screens out known unnecessary device files. These include:

mouse pty pts inet pri sxt mvw tp dsk/f[01]* rdsk/f[01]*
unneeded tty devices

Additionally, it screens for a total number of device files exceeding 900. Large installations or those with multiple serial port boards can have an unusually large number of device files in the /dev directory. If this is the case, an error message is generated, and you are allowed to either abort or continue.

You can temporarily get around this problem by editing the file /tmp/dev_files from another session and deleting device names that are not necessary. A more permanent solution is to edit the /etc/airbag script and change the EXCL_DEVICES environment variable to include the name of the devices you want to exclude. For example, if you want to exclude all the devices associated with FacetTerm, use:

EXCL_DEVICES=''/\/facet/d;''

Don't forget the backslash before the second forward slash. You can keep appending multiple patterns after the semicolon.

6.5.5 Files Added Conditionally  Back To Top

If you have a copy of the Universal Fdisk Partition Editor, UFDISK, this is automatically added space permitting. The vi editor is automatically included on the diskette as well. If there is not enough room for vi then the ed editor is included. The tape utility is added when you answer y to the question ``Run special commands before starting backup?''. The tape utility is useful for setting variable or fixed block mode for a SCSI tape drive. The tadpole file transfer utility is added to support the Remote Modem Support Feature.

6.5.6 Adding Custom Files to the A2 Diskette  Back To Top

You can have custom or personal binary files added to the A2 diskette by including the full pathnames (one per line) in the file /usr/cactus/airbag/A2aux.files. The Air-Bag diskette generator uses this list to add files (providing there is enough space) to the A2 diskette after all other important files have been copied over.

6.5.7 Completion  Back To Top

At the completion of the creation of the Air-Bag diskettes, you are notified if any errors were encountered, and if they were, you have an opportunity to view these on the screen. Some may be serious and others are simply warning messages. A copy of the most recent errors are kept in the file /tmp/airbag.err. You can later refer to this at your convenience.

Regardless of whether or not error messages were encountered, you must test out the diskettes! Never, never assume that they will work perfectly without testing! Your data is too valuable to risk.

You should always create a spare A2 diskette. Sometimes, the original is accidentally corrupted during use. If it gets corrupted, you need the spare.

6.5.8 Adjusting Custom Environment Variables  Back To Top

The following section shows the variables that can be customized and the possible values they can take. There may be more variables supported in future versions, so take a look at the top 20 lines of the /etc/airbag script for the latest information.

DO_COLOR ={YES,NO,ALWAYS,NEVER}

This variable affects the question regarding ANSI console color support.

A1_FS={XENIX,AFS,EAFS}

Setting A1_FS to either AFS and EAFS result in very slow booting. Change this only if you don't want to add XENIX support to kernel

COMPARE={NO,YES,VERBAL}

This is the option to bit-level verify the A1 floppy against files on the hard drive.

BTLD_ENABLE={NO,YES,NEVER,ALWAYS}

Support BTLD driver disks when creating the A1 diskette?

FLOPPY_TAPE={NO,YES,NEVER,ALWAYS} (NEVER for Open Server 5)

Do we ask the question regarding floppy based tape drive support?

PROMPT_TAPE_CMD={YES,NO,NEVER,ALWAYS}

Do we prompt for a tape command (such as ``tape reset'' or ``tape -a 1024'') when restoring? This is set during the installation and becomes permanent unless you reinstall. This is overridden (automatically set to ALWAYS) if you are generating diskettes for the BACKUP PROFESSIONAL.

VDISK_MANAGER={NO,YES}

Do we add support for the SCO Virtual Disk Manager? Please refer to the latest release notes for any changes to the above.

PORT={/dev/tty1a,/dev/tty2a}

The name of the default modem port for the Remote Modem Support Feature.

BAUD:={1200,2400,4800,9600,19200,38400,78600}

Default baud rate for the Remote Modem Support Feature.

RTAPE_SUPPORT={NO,YES}

REMOTE_HOST=''''

Do we support remote tape drives? If so the REMOTE_HOST variable should be set. See the release notes to determine if your version can actually support a remote host.

INCL_BOOTSTRING={keywords}

These are additional keywords that are automatically added to your default boot strings in /etc/default/boot when the A1 diskette is generated. These could be any one of the following: adapter=spad(0x340,11,0), cache=/d, mem=xxx,apm.exists=disable

EXCL_BOOTSTRING={keywords}

These are keywords that we definitely do not want added to the boot string in /etc/default/boot on the A1 diskette. This could be apm.check=verbose

KEYWORDS_AUX={keywords}

Additional keywords (besides internal defaults) to scan for in the default boot string.

EXCL_DEVICES={patterns}

These are patterns representing device files that should not be added to the A2 diskette. Set this if you exceed the maximum threshold of 900 device files. For example, to exclude /dev/facet0, use ``/\/facet/d;'' and to exclude /dev/xpr/xpr[0-9] use ``/\/xpr[0-9][0-9]*/d;'' which should appear directly after a forward slash in a pathname. Use this under guidance of technical support staff.

6.5.9 Booting from 2 Diskettes  Back To Top

Once you have successfully generated the A1 and A2 diskettes, you are ready for a test boot. First insert the A1 diskette into the floppy drive. Shut down your system the way you normally do. When the system is shut down, press the reset button on the computer. You see the boot prompt boot: and you should strike <RETURN>. After about 30 seconds, you are asked to remove the A1 diskette and insert the A2 diskette. You see configuration information on the console screen. Then the Air-Bag Main Menu appears on the console screen.

If you have a Northgate programmable keyboard, it may be necessary to be sure the macro mode is turned off before you boot. Otherwise, you may not see the prompt to put in diskette #2 (A2). The symptom of this is an error message that says:

Unable to load floating point emulator

6.6 Testing The A1 and A2 Diskettes  Back To Top

If you are using the Air-Bag diskettes for the first time, you must always test the integrity of the A1 and A2 diskettes by booting from them and selecting the following choice from the Airbag Main Menu.

9) Verify Air-Bag Integrity (Test Tape and Hard Drive Accessibility)

This tests both the accessibility of the tape drive and the hard disk. Without this test there is no way you can be sure the A1 and A2 diskettes will work in the event of a system crash. It is tragic to have a system crash and find your tape driveis not working correctly when booted under the Air-Bag diskettes. This can be a serious problem because some third party vendor's tape products do not work properly with the tape when booted from a floppy disk.

It is a good idea before testing the tape drive to set the WRITE PROTECT TAB on the tape. On 4mm and 8mm tapes, this is a slot that covers a notch. The slot is moved so the notch is left opened. On 1/4 inch cartridge tapes, this is a circular cylinder that is turned 1/2 turn with a small flat screw driver. Turn the cylinder so the arrow points to the word ``SAFE.'' If you do not want to verify the whole tape, but do want to make sure the drive is working, you can hit the <DEL> key while the tape is verifying. This takes you back to the Airbag Main Menu.

You should also test the Remote Modem Support Feature by testing the modem connectivity using the Airbag Utilities Menu.

If you have a floppy controller based tape drive and are not using OpenServer 5, you go through a three minute procedure of copying the Airbag Menu System to the free space on the hard disk. Don't worry. This does not overwrite any files on the hard disk. You then reboot to the hard disk based Airbag Menu System and proceed with testing the tape drive.

6.7 WHEN TO REGENERATE AIR-BAG DISKETTES  Back To Top

Anytime you reconfigure the tape (using ``mkdev tape'') and the kernel is rebuilt and relinked, you should regenerate the Air-Bag diskettes. If the SCSI ID of your tape drive is changed, you should regenerate the Air-Bag diskettes.

You need to regenerate the Air-Bag diskettes under the following circumstances:

1) You add a third party hard disk or tape driver
2) You change the sizes or number of filesystems
3) You change the mounting point of any filesystem
4) You add a DOS partition or any other partition
5) You add a bad track to the bad track table
6) You change anything using the divvy utility
7) After a restore when you selected NEW Hard Disk from the Airbag Menu System
*8) You add any new hardware card or change interrupts
*9) You relink the kernel and are not using /unix.airbag or /unix.N1
10) You change the SCSI ID of a tape drive or hard disk drive

* means optional, recommended but not required

6.8 JUMP-STARTING OPENSERVER 5  Back To Top

A jump-start option exists for OpenServer 5 that allows you to quickly boot to the Airbag Menu System. After booting the computer with the boot: prompt displayed on the screen, you type in the word airbag and the computer promptly boots to the Airbag Menu System running on a RAM filesystem in memory. You do not need the A1 or A2 diskettes for the jump-start feature. However, you must have a working primary hard drive with a valid bootstrap. All the features of the Airbag Menu System are active.

SCO OpenServer 5
boot: airbag

Here are some other options.

6.9 THE AIR-BAG MAIN MENU SYSTEM  Back To Top

Air-Bag is designed to be used from a comprehensive menu system called the Airbag Menu System. From the Airbag Main Menu, there are two additional menus: they are the Airbag Utilities Menu and the Airbag UNIX Expert's Menu. In this section we will describe the features of the Airbag Main Menu itself.

FIGURE 13. Air-Bag Main Menu System

From the Airbag Main Menu you can immediately start the restoration of a UNIX system, either partially or entirely. There are four modes of restoration, as follows:

•  
  • Root Filesystem Only
  • Fully Automated Restore to the Same Hard Disk
  • Interactive Restore to the Same Hard Disk
  • Interactive Restore to a New Hard Disk.

6.9.1 Root Filesystem Only Restore  Back To Top

The Root Filesystem Only Restore feature is designed to restore strictly the Root Filesystem, and saves you from having to perform an entire system restore. This should be used when you are certain that only the Root Filesystem data has been lost or corrupted. Keep in mind, this only restores the root filesystem. It does not restore the /stand filesystem under OpenServer 5. This may change in a future release, so read the release notes carefully.

There are two ways in which the Root Filesystem can be restored. First, you can perform a fresh install of the Root Filesystem. his destroys all previous data in the filesystem, remakes a fresh filesystem and restores it using the data on the tape. Use this when there has been structural damage to the integrity of the filesystem itself such as corrupted inodes, corrupted directories, or it cannot be cleaned properly with fsck. Second, you can perform a preserve install of the Root Filesystem. This installs data on top of the currently existing data. This could be useful if you don't have a recent backup of the Root Filesystem and have since installed additional drivers that are not present on your backup tape.

6.9.2 Fully Automated Restore  Back To Top

This feature is selected with the Airbag Main Menu choice:

2) RESTORE Backups to SAME Hard Disk (FULLY AUTOMATED)

The beauty of this particular feature is that a single keystroke followed by pressing the <RETURN> key eight times, fully and completely restores the entire system. The user need not know anything about hardware, software, or the operating system. The eight presses of the <RETURN> key allows default answers to questions regarding your particular tape drive. Please see the section entitled Tape Parameter Questions to see the actual questions asked.

The Fully Automated Restore is useful in the situation where you want to restore your data to the same hard disk with the same exact partitioning information. This is useful when your filesystem becomes corrupted by an aberrant program and is beyond the point of repair by the fsck utility. Another situation where this is useful is when there is the failure of a hardware component (i.e. hard disk controller card) but the hard disk is still intact. After replacing the defective component, you can restore the entire system automatically using this feature. Another situation where this is useful is when your hard disk crashes and is either a SCSI or IDE (Error free) and you replace it with an identical new or used hard disk.

There are other useful situations when this feature is used. These include:

1. Replacing an IDE hard disk with a SCSI disk or

2. Replacing a SCSI/IDE (Error free) with SCSI/IDE(Error free) of same/greater capacity

 

Keep in mind, that if you use this feature, you do not benefit from the increased capacity of a replacement hard disk. The configuration sizes are the same as the old hard disk.

If you use this feature on a system with more than one hard disk, and also need to rebuild the additional hard disks, you should prepare the additional hard disks first before using this selection. In this way, when the totally automated restore rebuilds the primary hard disk, all the other hard disks are prepared and will be mounted for restore of the Master backup.

After selecting this choice, you have one opportunity to abort. Once you select <RETURN>, the five phases start automatically. Thefive phases, if successful, appear as follows:

****************************************************************
** AUTOMATIC RESTORE of entire system **
****************************************************************

***Phase 1: Automatic FDISK...Success

***Phase 2: Automatic Badtrack...Success

***Phase 3: Automatic Boot strap...Success

***Phase 4: Automatic Divvy of filesystems...Success

***Phase 5: Automatic Restore...

MOUNTING FILESYSTEMS...
Mounting filesystem /dev/root on /mnt/...Success!!

The following filesystems are mounted...
/dev/root

Please insert Master Backup tape and strike <RETURN> when ready

The remaining questions are the same as those reviewed in the section of this manual entitled Tape Parameter Questions. In most cases, selecting the default choices suffice. After the tape information is restored, you are ready to boot from the hard disk and start using the system.

This feature takes into account the fact that you may be switching hard disk types from an ST-506 or IDE (w/ Bad Tracks) to a SCSI or IDE (Error free) hard disk. In these cases, the bad track table is zeroed out to reflect the error free nature of the new hard disk. When these cases arise, you see a ``*** Phase 2b: Clearing out Bad Tracks...'' message.

Also accounted for are situations where the hard disk tables and in-memory tables get out of synchronization. When this happens, corrective synchronization action takes place in Phase 4, and you are asked to reboot from the Air-Bag diskettes and try the Fully Automated Restore once more.

The Fully Automated Restore feature is primarily designed to recover from a crashed primary hard disk. It does not prepare or alter any of the add-on hard disks. Therefore, if your primary and secondary hard disks crashed, you should prepare the secondary disks first using the choice:

RESTORE Backups to NEW Partitions or NEW Hard Disk (Step by Step)

Prepare each of the add-on hard disks that have crashed or been replaced, then return to the Airbag Main Menu and run the Fully Automated Restore feature to prepare the primary hard disk and finally to restore the data. All previously prepared hard drives are mounted and participate in the automated restore.

You should not use the Fully Automated Restore if you are replacing an IDE (with Bad Tracks) with another IDE hard disk. This is because the replacement hard disk drive needs to have the bad tracks mapped out. The new bad tracks are different from the bad tracks on the original hard disk drive.

You should not use this feature when you are upgrading to a hard disk of a larger capacity. The extra capacity is not realized if you use this feature.

6.9.3 Restore to Same Hard Disk, Step by Step  Back To Top

The Restore to the Same Hard Disk (Step by Step) feature allows you to customize your entire system rebuild. This takes you through all the configuration screens that are available through the Air-Bag including the hard disk partition table and the filesystem division table.

6.9.3.1 Same or Expanded Partition, Corrupted Filesystem  Back To Top

If your filesystem is not intact, you will need to recreate it by dividing the blocks in the partition to the various filesystems. This is done through the divvy utility.

You select OPTION 3 on the main menu:

3) RESTORE Backups to SAME Hard Disk (Step by Step)

You are asked if the UNIX partition has been set to Active. If it has not, you are allowed to do so by using the fdisk utility. If you forget to make it active, it is done for you and a message stating the same is shown to the screen.

Next, you are shown the layout of your previous system. This includes the sizes of the filesystems in 1024 byte block and on which directories the filesystems were mounted. This is important information. If you are working at the console, you can view this information anytime by pressing <F-3>

Once you have looked at the previous filesystem setup, you enter the divvy utility. This allows you to divide the partition among different filesystems. If you selected all the defaults when you originally installed the system, all you have to do is hit <RETURN> several times and all the defaults are recreated again.

The divvy utility remakes all the filesystems on your partition which will, in effect, wipe out all data on the partition. If you arrived at this section of the menu by mistake, you are allowed to exit without harm by typing the <DEL> key. You are warned by a prompt when this is about to occur.

If you made special adjustments to the size of various filesystems when the partition was originally created, you have the opportunity to recreate this setup. You can see the original Divvy Table at anytime by pressing <F-4>. You should, when prompted, ask for a block-by-block control of the filesystem. You will see a start block and an ending block number for each filesystem. You should check to see that the difference between the two (which is the size of your filesystem) is at least as big as your original filesystem. You can make adjustments as needed using the menu prompts. If you get stuck and need either the previous filesystem setup information or the original Divvy Table information, you can see these at anytime by pressing <F-3> or <F-4> respectively.

The divvy utility remakes all the filesystems for you. Once this is done, you have the opportunity to restore your Master and Incremental tape backups.

All filesystems are restored at once from your logical filesystem contained on the Master backup. Air-Bag takes care of the necessary mounting and unmounting of these filesystems. You enter the tape device information including the name, block size, and capacity. The default values reflect the tape settings in effect when the A2 diskette was created. See TAPE PARAMETER QUESTIONS on page 159 for a detailed overview. You are prompted to enter the Master backup and start the restore operation.

After the Incremental tape backup finishes, your filesystems are restored. You go back to the Airbag Main Menu by pressing <RETURN> and then exit the Airbag Main Menu by selecting OPTION 99

99) EXIT and Shutdown

Then reboot from the hard disk. After the system comes up, you should remake your Air-Bag diskettes because your filesystem has changed. All the filesystem sizes have been changed and this information must be recorded on the new Air-Bag diskettes. b

6.9.3.2 Same Partition, Intact Filesystem  Back To Top

If you are restoring to the same partition, the procedure is fairly simple. There are two ways of doing this, depending on your situation. If your filesystem is intact (it can be cleaned by fsck) you can take a short cut and go straight to the Airbag Utilities Menu by selecting:

8) Utlities Menu

Once in the Airbag Utilities Menu, you select:

7) RESTORE Backup(s) Now

This allows you to restore your Master backup followed by your last Incremental tape backup. After this simple procedure, your system is up and running just as it was when your last Incremental tape backup was done.

6.9.4 Restoring to a New Partition or Hard Disk  Back To Top

The feature entitled Restore to a New Partition or New Hard Disk, provides total control over exactly how the new system is to be configured, even allowing you to take advantage of more room on a larger hard disk if it is available. If you need to expand the partition, adjust the size of DOS, or install on a new hard disk, you should select:

4) RESTORE Backups to NEW Partition or NEW Hard Disk (Step by Step)

This gives you the option of which hard disk to use. Your choices include the primary hard disk (main hard disk), the secondary hard disk or any other hard disk that was present at the times the A1 and A2 diskettes were created. As a rule of thumb, you should prepare your hard drives in reverse order. For example, if you have three hard drives, you should prepare the tertiary drive first, then the secondary and finally the primary. This allows you to use the totally automated feature instead of preparing the primary hard disk manually.

The restore procedure is similar to that in the previous section with the exception of prompts for setting your hard disk parameters if your hard disk is a nonstandard hard disk.

A nonstandard hard disk is one whose drive type isn't present in the ROM of the machine or one in which you wish to override the normal drive head/sector/cylinder mappings used by the hard disk controller. If the hard disk you have installed is nonstandard, you need the following information:

# cylinders
# heads
write reduce
write precompensation
ecc
control
landing zone cylinder #
# sectors per cylinder

You are asked the question:

[Default: no]
Do you wish to Enter/View HARD DISK PARAMETER information? (y/n)

If you have a standard hard disk, answer n and continue. Otherwise, answer y and enter the information. There are defaults provided for every field if you do not know all of the above information. The important items for partitioning the hard disk using fdisk are the cylinders, heads, and sectors per cylinder. The landing zone should be the last cylinder or the cylinder past the last cylinder (on some hard disks).

If you answered y to the question, a copy of the SCO boot instructions is automatically written to the Master Boot Block of your hard disk. This insures proper booting of the system with the nonstandard hard disk, when first turned on. If you answered n, the Master Boot Block may need to be initialized with booting instructions. You are given an opportunity to do this using the SCO boot instructions by answering the question and explanation:

Do you wish to put SCO Master Boot Assembly instructions
on the Master Boot Block?
 

Please note:
If this hard disk has never been initialized, answer 'y'.
 

If you have not created any other partitions, answer 'y'.
 

If you want to force write of SCO Boot instructions to Master Boot Block, answer 'y'
 

A 'y' answer does not overwrite the partitioning information, only the initial BOOT Code that is executed when the computer is turned on.
 

If you have already created other partitions using other operating
systems, answer 'n' unless you want to overwrite their BOOT Code
with that of SCO's.

Please respond(y/n)?

The previous explanations are self-explanatory. Even though this may seem somewhat complex, it does allow you to preserve the current boot instructions in he Master Boot Block which may have been placed there by another operating system.

Next you go into the fdisk utility to set the size of your partitions on this hard disk. Be sure that the UNIX partition is set to Active. This is a common mistake. Don't worry; if you forget, Air-Bag does it for you.

Once the disk is partitioned and the UNIX partition is set to Active, you can map out the bad tracks on the partition. You enter the badtrk utility and select the option to remap the entire partition. This may take some time (30 minutes or longer), depending on the size of your partition. Once complete, you can add any bad tracks to the list. Check the list that came with your hard disk. If there is a track that is on the list but was missed by the mapping, go ahead and add it to the list. This is far better than having that track go bad later on when there is valuable data on it.

If you indicated that you have a SCSI, OMTI, or IDE hard disk controller card, the scanning of the hard disk for bad tracks is skipped. In these cases, it is not needed, since bad track mapping is already handled by the controller.

You are asked how much room to save for future bad tracks in case later, others go bad. You should give yourself a buffer of 25% of the number you already have. If you have 20 bad tracks found, you should reserve 25. Usually the recommended default will suffice.

Once mapping the bad tracks is done, you are ready to divide the partition into filesystems. You are shown your old filesystem sizes and mount points. If you used the default filesystem setup, you can use this again by hitting <RETURN> several times. This is true if your new partition is larger than your old partition. If it is smaller, you should definitely look at the block-by-block control and make sure each filesystem has enough room. You can do this by subtracting the start block from the end block to give you the filesystem size. This size should be equal to or greater than the size of your old filesystem.

Once this is done, you restore your Master and Incremental tape backups. All filesystems are restored from the data contained on the backups. The Air-Bag programs take care of the necessary mounting and unmounting of these filesystems. You enter the tape device information including the name, block size, and capacity. The default values reflect the tape settings in effect when the A2 diskette was created. See TAPE PARAMETER QUESTIONS on page 159 for a detailed overview.

You are prompted to enter the Master backup, and then the Incremental tape backup. Once finished, your system is restored. You exit the Airbag Menu System and reboot from the hard disk, remember to remove the A2 diskette from the floppy drive before booting. After booting from the hard disk, you should remake the A1 and A2 diskettes because your filesystem has changed.

6.9.5 Modem Technical Assistance Service  Back To Top

The Remote Modem Support Feature is designed to allow complete remote support of your crashed system via modem. The menu interface that is typically seen by the user at the console can be seen by a remote technician thousands of miles away. The remote technician can take full control of the system and rebuild your crashed system.

This allows your dealer to troubleshoot or rebuild your crashed or damaged system. Additionally, technical staff at Cactus or UniTrends can provide emergency assistance for a support fee. If there is a need to transfer additional utilities to your booted system, this can be accomplished since the receiving portion of the kermit protocol (tadpole) is automatically installed as part of the Remote Modem Support Feature. This allows full support without the hassle of on-site travel.

In order to setup the Remote Modem Support Feature, several items should be in place when you generate your Air-Bag diskettes.

1. You need to have a HAYES compatible modem attached to either serial port tty1a or tty2a (same as COM1 and COM2).

2. If you are using a serial mouse, you must know to which port (tty1a or tty2a) it is attached since you cannot use this for your modem port.

3. Make sure the serial drivers are installed in the kernel. Run ``hwconfig -h'' to see if the serial drivers are installed. If not, run ``mkdev serial''.

4. Check the value of the PORT environment variable in the /etc/airbag file. It should be set to the name of the modem port (/dev/tty1a or /dev/tty2a).

5. Check the value of the BAUD environment variable in the /etc/airbag file. This should be set to the default baud rate of your modem.

Keep in mind that the default values can be changed even after you have booted from the Air-Bag diskettes. However, it is better to have them match the correct settings initially since in a crash situation it is an additional hassle to fiddle with these settings in order to find the correct one. Also the system is down and lost down time is lost money.

Be aware of this limitation: You cannot use the N1 kernel and expect it to work with the Remote Modem Support Feature. This is due to the fact that the N1 kernel does not include serial port drivers. Thus, the native N1 kernel or the A1 diskette generated using the N1 kernel will not work with the Remote Modem Support Feature.

You should test the Remote Modem Support Feature when you test your Air-Bag diskettes. After you boot from the diskettes and test accessibility of your tape drive, select the modem support feature from the main menu. An example of this menu is shown below:

FIGURE 14. Modem Support Menu

The settings for the modem port and baud rate are shown at all times. Make sure these are correct and then enable the modem using menu OPTION 1. While doing this, look at the lights on the modem and make sure they flicker. You should see the receive lights flash and then the TR light on the modem should stay on. When you disable the modem this light should go off. You should also see the AA light go on if it were not already.

With the modem enabled, you should then have someone dial into your system to make sure they can connect and that they see a shell prompt that appears like this:

Modem>

The remote user then either types the word cactus or menu and they should see the full Airbag Main Menu. If the screen is cluttered with control characters, then the remote site is probably not using an ansi terminal. In this case, the remote user exits the main menu, and runs the following from the prompt:

Modem> TERM=dumb; export TERM; menu

All the features including the UNIX Expert's Menu are available to the remote technician.

If there is not a remote user available to test your modem, you should at least dial the modem telephone number manually and listen for an answer. The modem should answer on the second ring. If you hear the typical modem handshaking squeal, at least you know that the basic physical connection to the modem and modem drivers are intact.

When using the Remote Modem Support Feature, keep in mind that you cannot change the baud rate or port device name while the modem is enabled. You must first disable the modem from the menu, then change the baud rate or port device name.

The remote technician works independently of the user at the console and has his own private session. Therefore, the console user should not be using the Airbag Menu System while the remote technician is logged into the computer. The console user can disable the port while the remote technician is logged on and this logs the remote technician off of the computer.

6.9.6 Additional Main Menu Features  Back To Top

In addition to the restoration methods described above, the following features are also available from the Airbag Main Menu, and appear numbered as follows:

5) Change Default Controller Type.

This option displays the default hard disk controller type and allows you to change it here on-the-fly. See SWITCHING HARD DISK CONTROLLER TYPES on page 162 for details.

7) Invoke the UNIX Shell.

This option invokes a UNIX Bourne shell session. To return to the Airbag Menu System when finished, type exit from within the UNIX shell.

8) Utilities Menu.

This takes you directly to the Airbag Utilities Menu, which is full of useful utilities and options, as well as providing entry to the Airbag UNIX Expert's Menu.

9) Verify AIR-BAG Integrity.

This option is used when testing the Air-Bag diskettes immediately after creation. Anytime you remake yur A1 or A2 diskettes, you should boot from the Air-Bag diskettes and chose this option. It is very important to select this option so that Air-Bag can attempt to read from the tape drive as well as access the hard disk drive. Without this test, there is no way to be certain that Air-Bag is able to read the archive media and rebuild the system after a hard disk crash.

6.9.7 Airbag Utilities Menu Features  Back To Top

The Airbag Utilities Menu shown in Figure 15,allows you to repeat a part of the recovery procedure if something failed. In general, you never need to enter this section of the Airbag Menu System. It is only designed for experts or for use during technical support calls.

The following is a complete list of the available utilities with a brief description of their purpose. They are presented in the same order as they appear in the Airbag Utilities Menu.

FIGURE 15. Airbag Utilities Menu

Set up DOS/UNIX Partitions on Hard Disk: The fdisk utility is used to partition the hard disk with operating systems. With every new hard disk, you must first decide how many different operating systems you want to configure it for, and then run fdisk to map out the space allocated for each OS.

Map out Active Partition Bad Blocks: The Airbag Utilities Menu allows you to view the bad tracks found on your previous hard disk. This is useful if the map is destroyed and you want to quickly key in the bad tracks by hand rather than doing a slow scan of the hard disk for bad tracks. Assuming there are no new interval bad tracks, this saves considerable time in restoring. This is the bad track utility which actually scans the hard disk for bad spots on the media itself, and then compiles this information into a table for future reference. Once a bad track is found, it is flagged so that it is not written to in the future.

Make Filesystems on Active Partition: The divvy utility divides the UNIX partition that was created with fdisk into UNIX filesystems. Additionally, divvy reserves space for UNIX itself to work in, and this area is known as swap space. For more information on the divvy utility consult your SCO System Administrator's Guide.

See how Original Filesystems were set up: This option shows you the way the original filesystems were structured at the time of A1 and A2 diskette creation. This should not be confused with the actual Divvy Table which can be viewed by invoking OPTION 11. This original information is important for future reference, especially if you are creating the same filesystems when rebuilding the system.

Put Boot Strap on Hard Disk: The information necessary to make the primary hard disk bootable can be put on the primary hard disk with a single keystroke. This option puts the SCO Bootstrap on the primary hard disk which gives it the ability to boot and load the UNIX kernel. If there is a Bootstrap already present on the hard disk, it informs you of this. If there is a Bootstrap which is other than SCO's, you are prompted to overwrite it or not. For instance, if you are converting a dedicated DOS or Windows hard drive to UNIX, you have a DOS or Windows Bootstrap. If your system was set up with the intention of booting from a number of operating systems, you may have a third party Bootstrap, which can be preserved by answering no to this question, and returning to the menu.

Select Hard Drive: This feature allows you to prepare single or multiple hard disks for restoration of data. You are given the opportunity to prepare any hard disk on the system.

Restore Backups Immediately: In some cases, you will have prepared a hard disk for restoration of backup data, but aborted before actually completing the restore from the archive media. This option allows you to restore archived data assuming the hard disks have been prepped.

Enter special HARD DISK PARAMETER INFORMATION: If you have a nonstandard hard disk, you need to gather additional information about your hard disk. See Restoring to a New Partition or Hard Disk on page 146 for the additional information you need.

View Controllers/Devices Recognized by the Kernel: This option allows you to see a list of devices/controllers that are recognized by your currently configured UNIX kernel. Use this option to tell if a particular SCSI hard disk controller card is recognized by the kernel or not.

View Saved List of Bad Tracks: By selecting this option, you are shown the current table of bad tracks for the hard disk you select. This option does not run the bad track utility, but you view the currently flagged bad tracks

View Saved Divvy Table: You can view the original Divvy Table of the primary hard disk. This is used for reference or used to look up the size of the swap partition. This option displays the original Divvy Table that was present on the system when the Air-Bag diskettes were created. This shows all the UNIX filesystems and their sizes as well as the swap and other raw partitions. This option does not actually run the divvy utility, as you are in a viewing mode only. You can also see the original Divvy Table at anytime by pressing <F-4>.

UNIX Expert's Menu: This option takes you directly to the Airbag UNIX Expert's Menu. The novice is cautioned against experimentation on a LIVE working system, as these utilities are extraordinarily powerful tools.

6.9.8 Airbag UNIX Expert's Menu Features  Back To Top

This chapter discusses the features of the Airbag UNIX Expert's Menu. They are presented in the same order as they appear in the menu seen in Figure 16.

FIGURE 16. AIR-BAG Expert's Menu

 

1) Diagnostic/Confidence Tests: This option runs a series of tests on the primary hard disk to determine if the hard disk is setup properly. The diagnostic tests appear as follows:

Check Partition Table in MasterBoot Block...Okay!!

Check MasterBoot Assembly Instructions Check Primary BootStrap...Okay!!

Check Secondary BootStrap...Okay!!

Check Divvy Table...Okay!!

Check BadTrack Table...Okay!!

2) Fix Damaged Super Block: This option is used when you get an error such as ``PANIC - srmount failure.'' The feature allows you to fix the corruption in the Super Block which is creating this problem. After the corruption has been fixed, you are prompted to clean the filesystem. Answer y (yes) if this is the only problem you are fixing. Answer n (no) if there are other problems, and then run the fsck utility after all the filesystem issues have been properly addressed.

3) Put Saved Divvy Table on Active Partition: This option actually replaces the current (and probably corrupt) Divvy Table with a copy of the original Divvy Table. This is especially useful when the Divvy Table is corrupted, or if the data for the filesystem sizes is lost or forgotten.

4) Put Saved Bad Track Table on Active Partition: The feature allows you to fix a damaged or corrupted Bad Track table with a copy of the original data.

5) Put Saved Critical Inodes on Root Filesystem: This option is very useful when there is low-level corruption in the root filesystem only. By invoking this option, the critical Root Inodes are immediately restored. You can use this if you see the fatal error message on booting:

Root Inode(Inode=2) not a directory or corrupt.

If this option does not fix corruption in the root filesystem, it may be necessary to restore it completely using OPTION 1 from the Airbag Main Menu.

6) Clear Master Boot Block: This option actually resets the entire fdisk partition table back to zero for all partitions. It is usually invoked when the Master Boot Block has become severely corrupted, and before selecting the next menu choice (OPTION 7). Air-Bag automatically saves a copy of your original healthy Master Boot Block for just such a situation.

7) Put Saved Master Boot Block on Main Hard Disk: If the Master Boot Block becomes corrupted or erased, this feature allows you to restore it instantly from the saved copy.

8) Zero-out Master Boot Assembly Instructions/Hard Disk Parameters: It may be necessary to zero out old assembly instructions that are no longer valid in the Master Boot Block. This is required, for example, when an old set of instructions is longer than a new set. Simply stamping the new set to the Master Boot Block will not overwrite the tail end of the old instructions. Sometimes this can lead to confusing behavior that is fixed by this option.

9) Overwrite Master Boot Assembly with SCO Version: If you have a nonstandard hard disk, you may need to initialize it with the SCO booting instructions. In some cases, there is a third party Bootstrap present on the system, either from a third party utility or another operating system installation. This option allows you to use SCO's Master Boot Assembly code to overwrite the Master Boot block, to ensure proper booting.

10) Make Main Hard Drive Bootable: This option is used in the case where the hard disk simply will not boot properly. By selecting this feature, you are essentially rebuilding the Boot Blocks exactly as they were at the time of the Air-Bag diskette creation. Furthermore, the Air-Bag will check the hard disk thoroughly before it is satisfied that the drive is bootable. Air-Bag checks for missing critical boot, inittab, and device files, and will actually repair them if possible. After this, it cleans the filesystem, and allows you to boot off the hard disk. If it is not possible to repair critical boot files, Air-Bag will notify you regarding exactly which files need to be replaced.

11) Delete Filesystems from the /etc/AIRBAG.MOUNTS list: This option is used in the case where a filesystem was mounted at the time of the A2 diskette creation, but it is not to be mounted when the system is rebuilt. When merging two filesystems into one, use this feature to delete the redundant filesystem.

12) Restore Saved CMOS Settings: This feature allows you to immediately restore the system CMOS settings to the most currently saved version made at the time of the A2 diskette creation. This is especially useful when the CMOS settings are blown away by a power surge, brown out, or dead batteries.

6.10 SPECIAL ACTION FOR FLOPPY CONTROLLER BASED TAPE DECKS  Back To Top

If you answer y to the question regarding floppy based tape drives during the creation of the A1 diskette, Air-Bag takes special action. This applies only to older SCO UNIX systems (3.2.2 and 3.2.4) and not to SCO OpenServer 5. Air-Bag proceeds normally until you are ready to access this tape drive. The tape is accessed either when you are ready to restore data or if you need to verify data on the tape drive.

At that point, the Airbag Menu System is copied to free space on the hard disk and you reboot and continue the restore or verify where you left off. When you reboot for the second time, you must type in the word again in order to access the newly copied Airbag Menu System. You type in this word when you see the boot: prompt when booting. Specific instructions are given to ensure there is no mistake about what to do.

The process of copying the Airbag Menu System to the hard disk takes an additional three minutes. This also occurs when you select the option from the main menu to verify the tape.

Keep in mind that when you are booted from the hard disk, a special message informing you of such displays on the Airbag Main Menu. Even though the Airbag Main Menu is presented after the tape restore/verify is completed, there are limited choices available. For example, you cannot setup a new hard disk when you are booted from the hard disk. If you select a choice from the Airbag Main Menu that is not allowed, you are informed of this fact; you are then asked to reboot from the
Air-Bag diskettes and select the same choice.

Restrictions do not apply to the Airbag Utilities Menu, but you should not enter this unless you know what you are doing and are aware that you are booted from the hard disk. You should never try to reconfigure the hard disk when booted from the hard disk. However, you can lay down the boot straps or restore data.

6.11 MERGING TWO FILESYSTEMS INTO ONE  Back To Top

If you have created the Air-Bag diskettes under a two filesystem configuration (/dev/root and /dev/u mounted on /usr) and you are converting to a one filesystem configuration, you can do this using one of two methods. You can use the Airbag UNIX Expert's Menu choice

Delete a Filesystem from the /etc/AIRBAG.MOUNTS list

or you can perform the task manually using the vi editor. The menu method is easier and less prone to error. The file /etc/AIRBAG.MOUNTS contains the filesystems that are expected to be mounted before doing the restore from the tape. If you are using the manual method, delete the line containing the filesystem that is no longer in use under the new configuration. For the case above, delete the entire line containing /dev/u. Do not make any other modifications to the file or the restore utilities will fail.

Once this file has been changed, you can select:

3) RESTORE Backups to SAME Hard Disk (Step by Step)

As long as you do not decrease the size of the UNIX partition, and you recreate the filesystems using divvy, there will be plenty of room in the single filesystem for both of the previous filesystems' data.

When entering the divvy utility, you need to request block-by-block control. You then delete one filesystem and add these blocks to the other filesystem. Do not, under any circumstances, change the physical location in the Divvy Table of the remaining filesystem. In other words if you are merging filesystems /dev/u and /dev/fs2 into /dev/u, do not change the physical location of the /dev/u filesystem in the Divvy Table. If you do, the restore process appears to work properly, but when you boot from the hard disk, the merged filesystem is not recognized. It will not mount, and chaos results.

When you boot from the hard disk after restoring data, go into SINGLE USER MODE and edit the file /etc/default/filesys. You need to delete any references to the previously deleted filesystem that are in this file.

6.12 SPLITTING ONE FILESYSTEM INTO TWO  Back To Top

Likewise, when converting from a single filesystem configuration to a dual filesystem configuration, you must add the entry for the second filesystem to the /etc/AIRBAG.MOUNTS file on the A2 diskette. The line must have five tokens, and one must be the word ``total''. For the filesystem /dev/u mounted on /usr use:

/dev/u /usr 54389 total blocks

The header of the file /etc/AIRBAG.MOUNTS should be left untouched. For convenience, copy the line representing the single filesystem (/dev/root):

/dev/root / 41990 total blocks

and change each of the five tokens to the appropriate name. The second token represents the name of the directory on which this filesystem is mounted and must be accurate. Air-Bag uses this when restoring data to the filesystem. The finished output looks like this:

FILE MOUNTED SIZE
SYSTEM ON (1024 byte blocks)
====================================================
/dev/root / 41990 total blocks
/dev/u /usr 54389 total blocks
 

The number representing the total blocks doesn't have to be exact. It serves only as a reminder to you when using the divvy utility in deciding how to allocate the blocks among the filesystems.

Once this is done, select the following choice from the Airbag Main Menu:

3) RESTORE Backups to SAME Hard Disk (Step by Step)

When you boot from the hard disk after restoring data, go into SINGLE USER MODE and edit the file /etc/default/filesys. You need to add references for the new filesystem. Follow the examples given.

6.13 EXCLUDING A SECONDARY HARD DISK   Back To Top

If only the primary hard disk of a two hard disk system was corrupted during a crash, Air-Bag allows you to restore the information to this hard disk only and leave the secondary disks alone. This is done by excluding the directory on which each secondary disk is usually mounted.

For example, let us suppose the secondary hard disk /dev/rhd10 is usually mounted on /usr2 on the root filesystem. Let us further suppose that the primary hard disk crashed and was replaced by a new one. You select the choice:

4) RESTORE Backups to NEW Partition or NEW Hard Disk

from the Airbag Main Menu and proceed through the steps as described above. Then when ready to restore the data, you are prompted to enter the names of any directories that you wish to exclude. You want to exclude the directory ./usr2 (or usr2 if using absolute pathname style of backups) since this is the directory on which the secondary hard disk is mounted. You see the following prompt:

[Default: no]
Do you wish to exclude any directories during restore? (y/n)

You should answer y (yes) to this question. Then you see:

Enter directory name prefaced with a period
 

Example: ./u
Example: ./tmp
Enter :

You then type ./usr2 and your input is confirmed with the following:

Excluding==> ./usr2
 

6.14 ADDING A HARD DISK  Back To Top

6.15 TAPE PARAMETER QUESTIONS  Back To Top

Whether you are restoring to a new hard disk, the same hard disk, or expanding a partition, the final phase involves loading your tape and restoring the data. You are asked several questions regarding the tape drive and the restore procedure. The default values for the tape information (name, block size, capacity) is the same as your primary backup tape device in use when the A2 diskette was created.

The following questions appear:

Please insert Master Backup tape and strike <RETURN> when ready
 

Type in the name of the tape device if other than '/dev/rct0':
 

Type in the tape blocking factor if other than 120:
 

[Default: no]
Does your Master Backup span multiple tape volumes? (y/n)

[Default: yes]
Do you wish to over-write a file if it already exists? (y/n)
 

[Default: no]
Do you wish to exclude any directories during restore? (y/n)
Enter directory name prefaced with a period
 

Example: ./u
Example: ./tmp
Enter :

Strike <RETURN> to continue

Please note that the tape device name must be the same name as that under which the data was backed up. If you backed up data to a 150 Mb tape under the name of /dev/rct0, it cannot be restored under the name of /dev/erct0. The two drivers use incompatible formats.

The name that you enter is tested to make sure that it is a valid tape device (character device) in the /dev directory. If you misspell the name, you are given repeated chances to enter it correctly. The tape blocking factor used must be the same as was used when the data was backed up. In most cases this is 120.

Additionally, if you are using 4mm or 8mm SCSI tape drives, the low-level SCSI block size must be the same as it was when the data was backed up. The low-level SCSI block size is different from the block factor and is related to the lowest level recording size used for the tape. It is specified in bytes and is set with the /usr/bin/tape utility. Generally speaking, if you use the system defaults for all your backups, you do not need to worry about the low-level SCSI block size. Otherwise, you can have the /usr/bin/tape utility placed onto the A2 diskette during its creation by answering y to the question regarding this. If you chose to do this, just before the restore commences, you have the opportunity to issue any tape command using the /usr/bin/tape utility. This is useful for skipping to different data sets or tape partitions if you have organized backups in this way.

If your Master backup tape set extends over multiple tape volumes you are asked to enter the size of the volumes in megabytes. The default capacity is taken from the tape library in the /etc/default directory and in most cases should be accepted as is. This size must be the same as the volume size used in the original command to back up the data. For a 550 megabyte tape cartridge, this value is 550. You cannot enter fractions. If you are not sure, it is better to overestimate the volume size than underestimate it.

If you want to avoid overwriting files that are already present on a filesystem, you have this choice. The default, however, is to overwrite all existing files with the data from the backup. If you choose not to overwrite existing files, the Non-Destructive restore feature is invoked and directory permissions are not changed. This means some directory permissions may not be restored (any directory path created to place a file or any directory that is already present), so overwrite the files unless there is a specific reason otherwise.

If you choose not to overwrite files and you have both a Master and Incremental backup, you must restore the Incremental backup first, then the Master. This is so you have the latest version of the files. This is the opposite of the prompts and the order of prompting has not been adjusted to take this into account. Therefore, when you are asked for the Master backup, put in the Incremental backup. When you are asked to restore from the Incremental, put in the Master backup. (Think for a moment about this until it makes sense to you.).

6.16 EXCLUDING DIRECTORIES  Back To Top

Occasionally, you need to exclude certain directories during a restore. Usually, this is a directory on which a secondary hard disk is mounted. Be aware that any directory that is excluded is automatically created as an empty directory. However, the permissions are set to full access (read-write-execute for owner, group, and others). Once booted from the primary hard disk, you may have to manually set the permissions if they are different than full access.

For example, if the directory ./tmp is excluded during a restore, you still want it to exist as an empty directory. Otherwise, application programs attempting to write data into the /tmp directory would encounter errors.

As many as 128 directories can be excluded. Individual files can also be excluded. You are prompted for each directory to exclude separately. Each choice is confirmed, and the entry is added to the total list of directories to be excluded.

Finally, the actual command that is used to restore the data is displayed. This is shown for those who wish to see exactly what command is being issued, so that if a problem is encountered there is a tangible basis for solving it. The data is then restored from the Master backup. The status of this restore is displayed on the screen. If this fails, repeated chances to try again are given. All questions are repeated so that opportunities exist to make corrections.

Following a successful restore of the Master backup, you are prompted for the Incremental backup. If this fails, repeated chances are given. Once the Incremental backup is restored successfully, exit the Airbag Main Menu and reboot from the primary hard disk.

6.17 SWITCHING HARD DISK CONTROLLER TYPES  Back To Top

If you have occasion to rebuild from a system crash on a computer that has a different hard disk type than the original, several adjustments must be made. First, you need to change the type of hard disk controller using a choice from the Airbag Main Menu.

Once this is done, you may or may not be able to access the hard disk. If the drive is an ESDI, ST-506, or IDE drive, it can be accessed by most UNIX kernels. If the drive is a SCSI drive and the Air-Bag diskettes were created on a system without SCSI support, you can only access the SCSI hard disk if the original kernel has SCSI capability built into it. This requires that when the kernel is linked, the file /etc/conf/cf.d/mscsi contain the following line:

Sdsk 0 0 0

Alternatively, if you choose to use the UNIX N1 kernel, this has SCSI capability built-in by default.

In all cases, when going from one hard disk type to another you should select OPTION 4 from the Airbag Main Menu

RESTORE Backups to NEW Partition or NEW Hard Disk (Step by Step)

When you enter fdisk or UFDISK, you should delete the existing UNIX partition and then add it back. This makes the ncessary adjustments for different cylinder/head configurations in the Master Boot Block.

6.17.1 Adding SCSI Controller Support to your Kernel  Back To Top

This section explains how to configure the kernel to include SCSI capability. Once you do this, you can make an A1 diskette with the new UNIX kernel. This new kernel allows you to easily migrate from an ESDI/IDE/ST-506 hard disk to a SCSI hard disk system.

The files shown in Table 25 should have a Y as the letter in the second field.

The /etc/conf/cf.d/mscsi file must contain the following lines:

The above assumes you will be using two SCSI hard disks and a SCSI tape drive at ID 2. If your tape is at a different ID than 2, change the 2 in the line containing Stp to the different ID. If you only use one SCSI hard disk, then remove the second instance of Sdsk.

If you are using the same SCSI controller card at all times, you can replace the auto keyword with the specific name of your SCSI driver. Table 26 shows the driver names of the common SCSI controller cards.

If you have a SCSI driver supplied with the SCSI card, use this name instead. You will know this because the SCSI installation instructions require you to type the word link at the boot prompt and insert a BTLD (Boot Time Loadable Driver) diskette supplied by the manufacturer of the SCSI controller card. You can look in the file /etc/default/scsihas for a comprehensive list of driver code names.

In this case, install the BTLD diskette using custom. You can do this even if the SCSI card is not yet physically installed in the system. For example, if the name of the BTLD driver is alad, you install the alad BTLD using custom. Then make sure the /etc/conf/sdevice.d/alad file has a Y in the second field, and the file /etc/conf/cf.d/mscsi has an Sdsk entry. Then relink the kernel. You have now added support for this driver in the kernel.

When you link the kernel, you will see a warning message stating that auto is supported for only 1 device. This is okay since it is assumed that you will only be using one SCSI controller card. You can safely ignore this message if you are using only one SCSI controller card.

Once you have made the above changes, as the root user, type in the following command to relink the UNIX kernel:

cd /etc/conf/cf.d
./link_unix

You should see the kernel being rebuilt. Ignore the WARNING message about ``auto''. Answer y to the question about the newly built kernel booting by default. It is not necessary to rebuild the environment. Sometimes this can cause problems with serial port cards if you haven't adjusted the /etc/init.d section of the link kit properly.

Once the new kernel is built. You should reboot to be sure you can boot properly from the new UNIX kernel. The kernel now has the ability to recognize many types of SCSI controllers and hard disks. You should then regenerate an A1 diskette by typing /etc/airbag and following instructions. Again, you should always reboot from the A1 diskette to be sure that it works and that you can access the tape drive using the UNIX kernel on the A1 diskette.

6.18 GENERATING CRASH DISKETTES FROM A BACKUP TAPE  Back To Top

If your Air-Bag diskettes have gone bad, you can regenerate them from saved images on the Master backup tape. This usually requires a spare system. It is ideal for dealers who always maintain an in-house Master backup of a client's system. If the client's Air-Bag diskettes are lost or damaged, the images of these can be retrieved from the Master backup.

All this is done through the Airbag Menu System selecting OPTION 5 and OPTION 6. This makes a copy of the A1 and A2 diskettes and puts them in the files /usr/cactus/airbag/A1.Z and /usr/cactus/airbag/A2.Z respectively on the hard disk. The menu appears as follows:

If your Air-Bag diskettes go bad, you can restore the images to fresh diskettes using OPTION 6. The menus that follow these choices are self explanatory.

Do not use these features as a fall-back or excuse for not creating two A2 diskettes. You should always have a spare, as the A2 diskette can become corrupted with frequent use. And in a crash situation, you will not have a good hard disk to use

The features in OPTION 5 and OPTION 6 are primarily for support staff that support clients, or in situations where a spare machine is available to restore the A1 and A2 images and create fresh A1 and A2 diskettes.

6.19 SCO OpenServer 5 UNIQUE FEATURES  Back To Top

Air-Bag runs under SCO UNIX 3.2v2, 3.2v4 as well as SCO OpenServer 5. There are unique features enabled when it detects it is running under SCO OpenServer 5 and these are:

• Full RAM Disk support
• Support for all 7 filesystems types including XENIX, S51K, S52K, AFS, EAFS, DTFS and HTFS
• Support for 2.88 Mb floppy diskettes
• You can run the Airbag Menu System by typing airbag at the boot: system prompt without A1 or A2 diskettes. This is called the jump-start feature.
• Ability to handle very large UNIX kernels using a three phase compaction technique
• The time to generate the diskettes has been reduced from five minutes to two minutes and 40 seconds

6.20 ALTOS SCO UNIX SUPPORT  Back To Top

The Altos version of SCO UNIX has renamed devices that are not the same as SCO UNIX. Air-Bag simply links the Altos names to the names that SCO uses for the Air-Bag to work correctly.

This is done by running the script mkaltosair, which is done automatically when Altos UNIX is detected. Air-Bag detects you have an Altos System if the word ``Altos'' is in the output of the ``uname -a'' command. Alternatively, you can set the environment varible ALTOS_UNIX=YES in the /etc/airbag script.

The script mkaltosair is automatically run after the A2 diskette has been created while the A2 diskette is still in the floppy drive. You will see the message ``Making special adjustments for Altos UNIX...'' This script modifies the A2 diskette. The /etc/divvy utility is moved to /etc/divvy_real and a wrapper script is used in its place.

This script file mkaltosair comes with the distribution and is found in /usr/cactus/airbag. You do not need to invoke it under ordinary circumstances. It is invoked automatically by the /etc/airbag script when you generate the A2 diskette.

6.21 1024 CYLINDER BARRIER PROBLEM  Back To Top

This problem results from a root partition that extends across cylinder 1024 of the hard disk. There is a SERIOUS problem when restoring to systems with greater than 1024 cylinders in the root partition. During the boot process, it is MANDATORY that the files /boot and /unix reside within the first 1024 cylinders of the system. If they do not, you receive such error messages as:

Stage 1 boot failure (this means "/boot" could not be read)
or
Bad Magic Number (this means "/unix" could not be read)

During installation, the Air-Bag checks for this problem and notifies you in no uncertain terms that it is present. You are referred to a file called README.1024 for a detailed explanation of the problem and how to fix it.

The solution to this problem is to create two filesystems. The root filesystem /dev/root should be made so that all data in this filesystem is within the first 1024 cylinders of the hard disk. The second filesystem, usually /dev/u should be the remainder of the hard disk. This guarantees that the files /boot and /unix residing in the filesystem /dev/root are always within the first 1024 cylinders.

The file /unix gets moved or copied every time a new device driver is added or the kernel is relinked. It is important that no matter how the kernel is relinked, the data blocks that are occupied must reside within the first 1024 cylinders of the hard disk.

If you simply do not have time to make two filesystems, you can run the script supplied with this distribution called putfront. This script putfront makes sure that the files /boot and /unix are put near the front of your root directory. This insures that when you do a Master backup, these files are backed up relatively early. If they are backed up early, then the files are located near the front of the tape, and they will be restored relatively early. This insures that they are on the front of the hard disk and within the first 1024 cylinders. You should run putfront anytime the kernel is relinked or new drivers are installed that rebuild the kernel.

The script putfront is found in the installation directory for this product which is /usr/cactus/airbag (unless overridden during installation).

6.22 LIMITATIONS  Back To Top

There are certain limitations in switching from one hard disk controller card to another. The primary limitation is that when you restore all the files, the restored version of UNIX may not be able to recognize the hard disk controller. This is because it was previously running under another type of hard disk controller. Therefore having a customized ``pleura-potent'' UNIX kernel (/unix.airbag) that can boot up under any type of hard disk controller is valuable. You simply use the kernel to boot from the hard disk. Then you can relink the UNIX kernel to recognize the new type of controller card. See Adding SCSI Controller Support to your Kernel on page 162.

Additionally, if you are caught without a customized (/unix.airbag) there are ways to relink the UNIX kernel on the hard disk when booted from the Air-Bag diskettes. It is too detailed to discuss in these notes how this can be done, but it is possible.

6.23 TROUBLE-SHOOTING  Back To Top

6.23.1 Trouble Creating AIR-BAG Diskettes  Back To Top

If you encountered any errors during creation of the Air-Bag diskettes, the error messages are saved for you in the file /tmp/airbag.err. You can refer to this file after you exit for useful clues regarding what happened.

The most common problem encountered is that too many files exist in the /dev directory. In this case, you should temporarily move some of the files out of this directory.

6.23.2 What to do if your Kernel is too Large  Back To Top

This is not usually a problem under SCO OpenServer 5 because Air-Bag boots from a compressed kernel. It can be a problem under older versions of SCO UNIX (3.2v2,3.2v4). If the kernel is too large, you have two choices. First, you can choose to use the N1 UNIX kernel. This works in most cases, but not all. It will not work if you have a third party hard disk driver or a third party tape driver. It will not work in some cases where you had to hard-code the tape parameters into the kernel using the ``mkdev tape'' command. If the tape parameters you hard-coded deviate radically from the defaults by the tape manufacturer, then the N1 UNIX kernel probably won't work. In many cases, the N1 UNIX kernel can figure out all of your tape settings during boot. The only sure way to see if the N1 kernel will work is to try it, then boot from the Air-Bag diskettes and test out the tape drive.

The second choice is to customize the UNIX kernel using the link kit. When you are finished, copy the customized kernel to the file /unix.airbag. A simple way to do this is to temporarily disable some of the drivers from being linked into the kernel. Next relink the kernel. Then you can re-enable the previously disabled drivers. If you have a TCP/IP network installed, these drivers can be temporarily removed and te kernel relinked. The drivers that you can temporarily disable and the amount of kernel space you save (in 512 byte blocks) are shown in Table 27.

In order to temporarily disable a driver, go to the /etc/conf/sdevice.d directory. In it you see many files. To disable the nb driver, you copy nb to nb.bak for safety. Then with an editor change the second field from a Y to an N. When you are done, it should look something like this:

nb N 256 0 0 0 0 0 0 0

If the field is already an N then this driver is not included in the next link. Then go to the directory /etc/conf/cf.d and run the program called ./link_unix. This relinks a new kernel. If the link proceeds without error then you have a kernel that is substantially smaller in size. You should answer n to the question of whether you want the new kernel to boot by default. The new kernel is located in /etc/conf/cf.d/unix. You should copy this to /unix.airbag and then regenerate the A1 diskette. Once completed, remember to reverse the changes you previously made to the files in /etc/conf/sdevice.d/.

Another way to create a smaller kernel is to uninstall all the drivers that do not address the console, hard disk drives, and tape drives. This includes the network drivers (TCP/IP), scanners, serial port card drivers, fax drivers, and other drivers.

Most drivers are uninstalled by using the ``mkdev xxx'' command where xxx represents the name of the driver. Look in the directory /usr/lib/mkdev for a complete list of uninstallable device drivers.Common drivers that can be uninstalled include:

dda nfs mouse token dos parallel shl slip vpixld

Even though you can change the system requirements using the etc/conf/cf.d/configure utility, experience has shown that this does not result in a substantial size decrease in the physical kernel as it appears on the hard disk. However, it does decrease the amount of memory that the kernel uses when it loads.

6.23.3 Trouble Booting from the AIR-BAG Diskettes  Back To Top

Sometimes the creation process will proceed without errors but you are unable to boot from the Air-Bag diskettes. This usually is caused by corrupted or bad floppy diskettes. To verify the integrity of a floppy diskette, you should select choice

4) Verify Integrity of a Diskette

from the Airbag Main Menu.This checks both the low level media format and the high level filesystem integrity. If the filesystem needs to be cleaned it is done so automatically. If you see any error messages in Phase 1, then the floppy diskette media is bad. You should discard it immediately. A new floppy diskette should be used that has been formatted and verified.

The command to format and verify a floppy disk in the primary floppy drive is:

For 3.5 inch floppy diskettes:

format /dev/rfd096ds18

For 5.25 inch floppy disks:

format /dev/rfd096ds15

If you see error messages in Phase 2, then the filesystem on the floppy diskette is corrupt and beyond repair. You should regenerate a new diskette and then verify it again.

6.23.3.1 Error Message: /etc/emulator not found  Back To Top

This error message results if you do not change the floppy diskettes when prompted. The kernel is looking for the file called /etc/emulator, and this file exists ONLY on the second diskette. You should try again and watch the screen carefully for the prompt to change volumes.

Sometimes, if your keyboard is not responding properly, this may happen. The message to insert the A2 diskette flips by quickly and is automatically answered from the keyboard queue. Try turning your system completely off, then on again, and repeat the procedure.

6.23.3.2 Error Message: EEEEE  Back To Top

If you accidentally try to boot from the A2 instead of the A1 diskette, you will see EEEE... on your console screen accompanied by a grinding noise from your floppy drive. This occurs if you forget to remove the A2 in the floppy drive and then try to reboot. The screen fills up with E letters until you either open the floppy drive door or turn off the computer.

If you want to boot from the Air-Bag diskettes, put the A1 (boot diskette) in the floppy drive and press the reset button. If you want to boot from the hard disk, no floppy diskette should be in the floppy drive at all.

6.23.3.3 Running out of space on boot floppy  Back To Top

If you receive out of space messages referring to the A2 diskette, you can safely remove the following files from the diskette:

/bin/vi
/bin/ed
/usr/bin/tape
/usr/bootup/*
/bin/df

Some of these files may not be on the floppy diskette because there was not enough space when the diskette was created. However, if any of these files do exist, they can be safely removed without affecting the performance of the Air-Bag.

Obviously, you must do this procedure when booting from the hard disk and after mounting the floppy diskette on the hard disk using the command:

mount /dev/fd096ds18 /mnt (3.5 inch floppy)
or
mount /dev/fd096ds15 /mnt (5.25 inch floppy)

In some cases, you can run out of space because the free list on the floppy diskette was corrupted. In this case, when booted from the hard disk (UNIX or XENIX), put the floppy diskette in the floppy drive and type:

fsck /dev/fd096ds18 (3.5 inch floppy)
or
fsck /dev/fd096ds15 (5.25 inch floppy)

6.23.3.4 Kernel PANIC - out of SWAP space  Back To Top

This happens when you create the Air-Bag diskettes on a system with a larger amount of memory than the one on which you have just tried to boot. Most likely, the kernel being used is too big for the machine from which you are trying to boot.

You should remake a new A1diskette, but this time choose to put on the /unix.N1 kernel. This usually solves the problem. This kernel has a low memory requirement.

As a rule of thumb, you need at least 750 K to 1000 K of user memory. The amount you have can be seen from th initial message on the screen after boot. It is near the bottom of the screen, and appears as follows:

mem: total = 5760k, kernel = 4012k, user = 1748k

In the above example, there is a little under 6Mb total memory. The kernel occupies 4Mb, which leaves roughly 1.7Mb for the users. If there is not enough user memory, the kernel tries to swap out and has no place to go and then PANICS and aborts.

6.23.3.5 Menu doesn't appear to respond to keyboard input  Back To Top

Turn your computer off, then on again, and restart the procedure from scratch. Some keyboards can get programmed incorrectly on booting or can have the <NUMLOCK> set to on when it should be off. See the warning above regarding Northgate programmable keyboards.

6.23.3.6 Kernel PANIC: main - swapadd  Back To Top

This can happen when the kernel is trying to add the swap space and it cannot do it. If you receive this message, there is an incompatibility between the kernel you are using and the ram(8) memory device driver. You should go to the /usr/cactus/airbag/mkboot.dir directory. You see a file called def_boot.2alt. This is the alternate boot instructions that avoids using the ram(8) driver and thus solves this problem. Simply copy def_boot.2alt to def_boot.2vol after making a backup copy of def_boot.2vol. Then regenerate the A1 diskette and reboot and this problem is resolved.

6.23.3.7 PANIC - srmount failed  Back To Top

This means that when trying to mount the A2 diskette (filesystem diskette) a failure occurred. This is usually because of a corrupted A2 diskette or in fact you put a floppy diskette other than the A2 diskette in the drive. You should have made a spare copy of the A2 diskette. This can be used. Alternatively, you need to take the A2 diskette to a working computer with the Air-Bag installed and select OPTION "4) Verify Integrity of a Diskette" and see if it passes the verification.

6.23.3.8 Does not prompt for a second diskette  Back To Top

This can happen in two situations. First, you may have hit <return> twice by accident at the boot: prompt. It is possible that you hit <return> once but the keyboard sent two return codes to the computer. A second reason is that you typed in your own boot string and forgot to include the word prompt=message in the boot string. If you forget to include the word prompt=message in the boot string, the boot process assumes that there is only one diskette and does not prompt for a second diskette.

6.23.3.9 Hangs after H6  Back To Top

The system prompts for the second diskette and after it is inserted the normal kernel messages appear until H6. Then nothing happens. This can be caused by several problems. During the H6 phase of loading the kernel several things happen. These include opening the root device (floppy filesystem diskette A2), loading the superblock into memory and mounting the root filesystem. A hang here indicates a damaged, missing or incorrectly configured root device.

6.23.3.10 Hangs after kernel: i/o bufs = 600 k  Back To Top

This happens when a problem occurs in initialization of the A2 diskette. It happens if you accidentally press <RETURN> twice in response to the boot: prompt and you have a floating point unit already installed on the computer or a 486 CPU. If you do not have a floating point unit you see the message: cannot load floating point emulator.

Alternatively, if files in the /dev directory do not get copied over correctly this can happen. Also, if the file /etc/init or /bin/sulogin on the A2 diskette have been corrupted this can occur.

6.23.3.11 No space on floppy diskette unit 0, minor 64  Back To Top

This message displays after inserting the A2 diskette and keeps repeating itself so that nothing can be done. The Airbag Main Menu never appears. This means that the floppy diskette filesystem needs cleaning with fsck.

If you reboot, the A2 diskette automatically detects that it has run out of space the previous time, and cleans itself before loading the Airbag Menu System. In most cases, this fixes the space problem. If it does not, try taking the A2 floppy diskette to a working system and running the ``fsck /dev/install'' command.

Most often it occurs if the A2 diskette is removed prematurely from a booted system. Also, when shutting down from the booted A2 diskette, if you do not wait for the light to go out on the floppy diskette before removing it, this will happen.

It is also possible that the floppy diskette was corrupted during its last use and the free list was corrupted causing the diskette to appear to be out of room. Corruption can occur from a variety of factors. Please see the section below entitled A2 diskette keeps getting corrupted.

6.23.3.12 A2 Diskette keeps getting corrupted  Back To Top

When you are booted from the Air-Bag diskettes, strange error messages start to appear, then get worse and worse until finally the diskette is inoperable and you have to turn the computer off manually.

Since every 30 seconds the A2 diskette is updated from memory, any corruption of memory causes corruption on the A2 diskette. The most common cause of memory corruption is a serial port controller card with dual ported memory that is set for a conflicting memory address. This happens if you add a new serial port controller card and select a memory address that truly exists.

Alternatively, if new memory is added to the system and the new memory overlaps a previously existing address that the serial port card was using, memory corruption can occur. This is even worse on cache systems because the dual ported memory needs to be outside of the range that the memory cache unit is using.

Any piece of hardware that can corrupt memory can cause this problem. Many times when using the Air-Bag, a known problem is fixed and then an attempt to restore the system is done. If other problems exist, that are not known about, they can be reflected by repeated corruption of the A2 diskette. For example, a lightning strike or power surge causes damage to the hard disk drive and the hard disk is replaced. It may not be noticed that the hard disk controller card is also damaged. It may manifest itself by corrupting memory and thus the A2 diskette. So, when seeing a corrupted A2 diskette, also consider the original reason that one is using the Air-Bag.

The Paradise VGA controller card or a VGA card that uses the Western Digital VGA chip has a bug in it. It can cause corruption when writing to the floppy diskette or tape drive. The only way to get rid of the bug is to put the 16 bit VGA card in an 8 bit slot. This slows down the VGA screen some, but it is the only way to get around the problem.

6.23.3.13 Sdsk: Unrecoverable error xxxx...  Back To Top

When running from the A1 and A2 diskettes, the error message appears

Sdsk: Unrecoverable error xxxx...

This can occur if you replace a hard disk with a smaller hard disk and try to do the totally automated restore. Or it will happen if you go to the Airbag Utilities Menu and stamp the old Divvy Table on the hard disk. The errors occur because the kernel is trying to access disk sectors beyond the size of the smaller current hard disk. You should use the selection:

4) RESTORE to NEW Partition or New Hard Disk (Step by Step)

when you replace a hard disk. The other choices won't make much sense and result in the above error.

6.23.4.1 Hard Disk Not formatted Error Message  Back To Top

You receive this error message if you try to access the hard disk while booted from the Air-Bag diskettes. The message tells you to low-level format the hard disk.

This results when the hard disk is not recognized by the UNIX kernel. You should reboot to see if the kernel recognizes the controller card. If it is a SCSI controller card, you should see a line containing the name of the card something like:

adapter 0xfc00-0xfc60 10 - type=bhba unit=0 id=7

If you do not see this, you need to add SCSI support to the kernel for this controller card. See Adding SCSI Controller Support to your Kernel on page 162.

If you do see the controller card, it is likely you have a DPT controller. Change the controller type to DPT from the Airbag Main Menu. If the problem persists, you have to go to the Airbag Utilities Menu from the Airbag Main Menu and go through each choice manually. Sometimes, escaping to the shell and typing the following solves the problem:

dparam -w

If you see: %Sdsk just before the error message telling you to low-level format the hard drive, it means the hard disk controller is recognized by the UNIX kernel. However, the hard disk cannot be accessed. Check the controller configuration, consider formatting if not done, or manually do the recovery from the Airbag Utilities Menu.

6.23.4.2 cactus: cannot make pipe  Back To Top

This error message can occur if you use the SCO 3.2.4 N1 diskette and the Air-Bag A2 diskette. The reason for this is that the SCO N1 diskette uses a boot string that does not create a pipe device. Therefore, when the kernel messages initially come to the screen if you look closely you see:

pipedev=<none>

Any program that pipes the output of one program into the input of another will fail. The solution to this problem is to avoid the use of the SCO UNIX 3.2.4 N1 kernel; instead, create an A1 diskette from the /unix.N1 kernel. This is done using the Air-Bag diskette generator and you are automatically prompted to put in the N1 disk, copy the kernel, and then put it on an A1 diskette. This allows the functionality of the N1 kernel, without the headache of not being able to create pipes when booted from the Air-Bag diskettes.

Alternatively, you can get around this limitation if you do not mind typing in a long string at the boot prompt.

boot: fd(64)unix root=fd(64) swap=ram(8) pipe=ram(1) swplo=0 nswap=32 prompt=''Insert A2 floppy, press return''

6.23.4.3 divvy: could not open temporary node  Back To Top

This may occur when you do not have an active partition on the hard disk. Most often you have created a new UNIX partition and have forgotten to make it active. If there is no active partition, the divvy utility displays this message, because it does not have a clue as to which partition to divide into filesystems.

Air-Bag actually checks to make sure you have made an active partition. If you do not have an active partition, it reminds you that you forgot to make it active and makes the UNIX partition active for you.

6.23.4.4 divvy: could not make temporary character/block node  Back To Top

In this case, the divvy utility is having trouble making a temporary device file in the /dev directory. This could be because the /dev directory is full, there re no more inodes left on the floppy disk, the node file that it is trying to create is already present, or there is no longer write permission in the /dev directory.

6.23.4.5 divvy: could not get drive parameters  Back To Top

This may happen if your main drive /dev/rhd00 is not really a device file or perhaps is the wrong device file. Check the major and minor numbers using the ls -l command and be sure that they look similar to what follows:

crw------- 2 sysinfo sysinfo 1, 0 Jan 22 1994 /dev/rhd00
^^^----major number is here
^^^-----minor number is here

Also, do the same for /dev/rhd0a and you should see:

crw------- 2 sysinfo sysinfo 1, 47 Jan 22 1994 /dev/rhd0a

Air-Bag does some basic checks on these devices, but it cannot detect incorrect major or minor numbers.

6.23.4.6 Select Add-on HARD Disk and computer freezes up  Back To Top

This is a problem with the UNIX kernel. If you go to access a device such as a secondary hard disk, and the hard disk is not really present, the entire computer freezes up. There is nothing we know of that you can do at this point other than manually reboot the computer.

Even if you do not have an actual secondary hard disk drive installed, Air-Bag makes these device nodes on the A2 diskette for you. This is for flexibility and transportability of the Airbag Menu System. If you accidentally select the wrong choice of primary versus secondary hard disk, the computer can hang and you must reboot. Sometimes you may accidentally strike key 2 instead of 1 and inadvertently cause the computer to freeze.

6.23.4.7 Totally Automated Restore Hangs in Phase 4    Back To Top

Look in the file /tmp/phase4 to see the reason for the failure. One reason is that you are using the Air-Bag on a system with far less memory than was present when the Air-Bag diskettes were created (e.g. 8-12 Mb ram). There may be a problem with recreating the division table due to a very large filesystem or the 1024 cylinder barrier problem.

6.23.5 Trouble Restoring a System    Back To Top

6.23.5.1 Hard Disk Error Messages during a Restore  Back To Top

The error message you might see is:

HELP - extract large write error because No such file or directory
or
HELP - extract large write error because No space

If this happens, tere are two possibilities. First, you may have a defective hard disk or hard disk controller. You may have a bad cable that connects the two. Anytime you adjust cards in the computer, there is a chance that you could bump a card loose or bump a cable. If you see a pattern such as the same hard disk head number appearing in all the bad block error messages, this implies that you have had a head crash and one of the heads on the hard disk is defective. In this case, it is time to get a new hard disk.

Second, it may be that the hard disk bad tracks were not mapped out properly or were not mapped out at all. This can happen if you selected the wrong type of hard disk controller when the A2 diskette was created. Also, if you selected a type of hard disk controller that does not need bad track mapping (SCSI or IDE error free) but in reality, you have a hard disk controller that does need bad track mapping (IDE -with errors, ST-506, ESDI). In any of these cases, if you suspect a problem with bad disk blocks, you reboot from the Air-Bag diskettes, select the proper hard disk controller type from the main menu, and then select ``4) RESTORE Backups to NEW Partition or NEW Hard Disk (Step by Step)" from the main menu to reinstall everything.

Third, it may be that you have run out of space on the hard disk. Sometimes this is not evident from the initial error message, but it becomes evident when you see the following message:

WARNING: No space on device (1/40)

This message may flash repeatedly up the screen. The reasons for this problem are discussed extensively in the next section.

6.23.5.2 Running Out of Space  Back To Top

There are several reasons for running out of space. First, if you had virtual files on the hard disk and you did not back them up using the special virtual file features, when you restore them they take up more space after the restore than they did originally on the hard disk when being backed up. Second, if you adjusted the sizes of the filesystems, perhaps a mistake was made and the actual size was made smaller than you thought. Third, if you have several file systems, it is possible that a critical filesystem was not mounted before the restore process. In this case, all the data that was supposed to go into this filesystem goes into the root filesystem causing it to run out of space. Fourth, hardware problems can cause this since the filesystems were never created properly in the first place because of the hardware problems.

6.23.6 Trouble Booting once you have Restored Files  Back To Top

6.23.6.1 HDD Controller Failure  Back To Top

Your CMOS is set incorrectly and the system bios cannot recognize the hard disk controller. It is also possible the jumpers are incorrect or the cables connected incorrectly. IDE cables should be checked carefully making sure the red colored wire lines up with pin 1.

6.23.6.2 Error message: NO OS  Back To Top

This means that no valid operating system is found in the active partition. This is usually because the boot straps have not been copied to the hard disk properly. This most commonly occurs if you have selected restore from the same hard disk when, in fact, a new hard disk had been installed. Additionally, it can occur if you change the location of the active partition through the fdisk utility.

The solution to this problem is to simply go to the Airbag Utilities Menu and select the choice that specifies copying the boot strap to the hard disk. You do not need to re-restore files if you have already done so.

6.23.6.3 Error message: IO ERR  Back To Top

This happens when the cylinder and sector values in the Master Boot Block do not reflect your current hard disk. You should check your hard disk BIOS type that is configured in the SETUP software for the computer hardware. Once this is correct, you should use Cactus International's UFDISK product to fix the problem. Alternatively, you can delete the active partition completely and then remake it. When this is done, the cylinder and sector offsets are recalculated.

6.23.6.4 Stage 1 boot failure: error loading /boot  Back To Top

This is probably the most common problem that you run into if there is going to be a problem. It means that the bootstrap programs (which have loaded successfully) are having trouble loading the program called /boot (which loads the kernel). There are many causes for this including:

• Incorrect disk geometry in CMOS
• Incorrect disk geometry in Master Boot Block
• /boot and /unix were never on the tape you restored from
• 1024 cylinder barrier problem
• Larger hard drive has different cylinder/head/sector map than previous drive
• Compaq System-Pro needs new EISA geometry stamped into CMOS
• The root or stand filesystem was corrupted
• Undetected synchronization error during the Totally Automated Restore

The exact reason for the failure appears just before the final error message. For example,

/boot not found
Stage 1 boot failure: error loading /boot

means the boot strap program cannot find the file called /boot in the root directory. Possible error messages and their meanings are shown in Table 28.

The first thing you should do is thoroughly review your disk CMOS settings and make sure they are correct. If you have a EISA system, check the disk geometry settings and if the new controller card is using a mapping scheme of which you are unaware.

Second, run the diagnostics and system confidence test by selecting OPTION 1 from the Airbag UNIX Expert's Menu. Alternatively, you can type the word diag from the Airbag Main Menu. This checks the fundamental integrity of the partitions and bootstraps.

Third, if under SCO OpenServer 5, make sure the /stand filesystem was actually restored. Make sure that the /stand filesystem is present on the hard drive by mounting
/dev/boot on /mnt while at the shell. You should see the files boot and unix and link, and they should all have a true size. If the Master backup was done by excluding all read-only mounted filesystems, /stand would have been excluded since it is normally mounted read-only. In this case, no files from /stand are on the Master backup tape.

In this case of an undetected synchronization error, you should start the entire crash recovery procedure over from scratch. You should try the totally automated recovery selection from the Airbag Main Menu again.

If you are not sure the CMOS settings are correct, you can restore them using the Airbag UNIX Expert's Menu. Then try booting again from the hard disk.

The 1024 cylinder barrier problem may occur if the /boot file is located on a cylinder number greater than 1024. Alternatively, part of the file may be located on a bad block that has not been mapped out. Please refer to the section of this manual entitled 1024 CYLINDER BARRIER PROBLEM.

If you cannot isolate the problem, repeat the restore again. If this fails, use the Airbag UNIX Expert's Menu and zap the Master Boot Block and restore using OPTION 2 from the Airbag Main Menu to restore to a new hard disk or new partition.

6.23.6.5 Kernel hangs at H6  Back To Top

If the computer hangs while attempting to boot and the last message is H6 this means that the hard disk controller cannot be initialized. Usually this happen if you switch hard disk controller types and the new type is not built into the kernel from which you are currently trying to boot.

Another possibility is that you have switched to a SCSI hard disk but forgot to go into the computer's setup routine and set the hard disk drive type to 0.

Alternatively, the files /dev/root or /dev/hd00 may be missing from the hard disk. You can fix this from the Airbag UNIX Expert's Menu using OPTION 10 ``Make Your Hard Disk Bootable''.

In the first case, since you have restored all files to the hard disk, there should be some alternative kernels present in the root directory. The kernels /unix.airbag and /unix.N1 may be present in the root directory.

You should try to reboot, but at the boot prompt type in the name of these alternative kernels. Do not include the leading slash.

If this does not work, you can boot from the A1 and A2 diskettes, mount the root filesystem /dev/root on /mnt. Then you can type in the command:

/mnt/bin/chroot /mnt /bin/sh

This makes the root of your /dev/root the main root directory. You can reset up your PATH variable, then change directories to /etc/conf/cf.d and relink the kernel adding in the drivers you need. The only thing you won't be able to do is use the floppy drive that had the A2 diskette in it. It will be similar to being booted from the hard disk.

Alternatively, you can boot from the A1 diskette, but instead of striking <RETURN> at the boot: prompt, type in hd. This uses the kernel from the A1 diskette, but otherwise will use the hard disk for the root filesystem and swap space and pipe device once the UNIX kernel is booted.

6.23.6.6 WARNING: No hd root controller  Back To Top

This means the kernel you are booting from does not recognize the hard disk controller card. You can temporarily boot from the Air-Bag diskettes by typing the word hd at the boot: prompt. This uses the kernel on the A1 diskette, but boots to the hard drive.

You then need to link the correct drivers into the kernel and try booting again from the hard drive. See Adding SCSI Controller Support to your Kernel on page 162.

6.23.6.7 Kernel Reports the Disk Drive has 65536 Heads  Back To Top

This is a bug in OpenServer 5 and has to do with getting the correct number of heads from the disk drive. This happens when the true number of heads is 128 or higher. Type in the word biosgeom at the boot: prompt to work around this problem.

6.24 ADVANCED TOPICS  Back To Top

6.24.1 Raw Partitions   Back To Top

Many high powered databases use raw partitions. These are areas of the hard disk reserved for exclusive use by database applications. The format of the data is known only to the exclusive programs using the raw partition. The format is not known to UNIX other than it is a large chunk of information. The UNIX operating system essentially ignores this section of the hard disk drive.

If the raw partition is on the primary hard disk (boot disk), it is accounted for during the Fully Automated Restore. The data in the raw partition is never touched by the Air-Bag. Therefore, unless your raw partition data was corrupted, it is intact when the Air-Bag has completed it's job. Any number of raw partitions on the main hard disk are supported.

It is assumed that you regularly make a special backup of the raw partiti data, in case it needs to be restored separately. This data is not backed up by the tar or cpio utilities. It can be backed up by any of the SUPER-TAR products such as CTAR or Lone-Tar. Also, a simple backup using the dd utility will suffice. The backup of this data is the responsibility of the System Administrator.

Raw partitions on secondary hard disks are not accounted for by the Air-Bag. You need to recreate these manually when the secondary hard disk is ``divided up'' into filesystem and raw partition divisions using the divvy utility.

6.24.2 Virtual Disk Manger Support  Back To Top

There are several things to point out.

First, AIR-BAG detects whether or not you are using the Virtual disk manager by looking for the file /etc/dktab. If it has a size, and you have /etc/dkconfig installed on your system, it is assumed you are actively using the Virtual Disk Manager.

Second, you should not use the AIR-BAG to reconfigure the sizes of a hard disk that contains both Virtual Disk Manager partitions and standard partitions.

The SCO 'divvy' program is intimately tied into the Virtual Disk Manager. This is good and bad. If you recreate a new setup using the AIR-BAG and then restore your system, the /etc/dktab reflects the old configuration, which may conflict with the new division table. Divisions that previously did not conflict, may now because of the different sizes. The hard disk you are replacing has mixed partitions, some with the Virtual Disk Manager and some with divvy. If the Virtual Disk Manager uses only entire drives, then you should not experience problems.

If you have mixed partitions and change the size of your hard disk, it is best just to copy /etc/dktab to /etc/dktab.bak after a fresh restore using the AIR-BAG. Then recreate /etc/dktab using the Virtual Disk Manager software under your SCO Desktop. Do NOT RECREATE a fresh AIR-BAG until you recreate a properly working /etc/dktab. Otherwise, the AIR-BAG will HANG when it tries to gather the divvy information. Divvy itself will not run properly and complains until the Virtual Disk Manager partitions and the divvy partitions are not conflicting.

Third, the current version has limited support for RAW Partitions that are also Virtual Disk Partitions. They are supported only if you do not exclude them from a RESTORE when booted from the AIR-BAG. Thus, you should not exclude Virtual Disk Partitions that are also RAW Partitions from a restore. The reason for this is the AIR-BAG creates fresh filesystems on all the Virtual Disk Partitions when you remake filesystems or do a RESTORE from the System Crash AIR-BAG. If you exclude a RAW Partition from the RESTORE and it is a Virtual Disk Partition, it will not have its original data on it and thus will need to be restored anyway.

Fourth, the Totally Automated selection can potentially remake ALL your Virtual Disk Partitions. Thus, be sure you have a good Master Backup of ALL Virtual Disk Partitions on all hard drives. In this particular case, it does not limit itself to Virtual Disk Partitions on the primary drive.

You will be prompted, however, to remake each filesystem. Thus pay particular attention to the name of the filesystem and if you want to remake it. Although this prompting feature was not in the original BETA for Virtual Disk Partitions, we have added the prompting feature at your request since many times you do not want to restore a Virtual Disk Partition that is on secondary and tertiary hard drives.

The consequence of this is there are a few more questions to answer when doing the Totally Automated Restore with Virtual Disk Manager Partitions. Most of you felt this was a good tradeoff