Technical article 126683

Under SCO OpenServer 6.0.0 and UnixWare 7.1.4, I can't access the USB DAT Tape Drive via the newly created Emergency Recovery CD (DRF) in order to do a system restore.

OpenServer 6.0.0UnixWare 7.1.4

Written · Last revised

Problem

I have a server that has a USB DAT Tape drive and CD-RW drive but no legacy 1.44 floppy Drive.

My "sdiconfig -l" output:
0:0,63,0: HBA     : (mpt,1)LSI Logic 1068 10414
  0,0,0: DISK    : SEAGATE ST336754SS      1503
1:0,2,0: HBA     : (ide,1) Generic IDE/ATAPI
  0,0,0: CDROM   : HL-DT-STDVDRAM GSA-H10N JH04 <-- DVD-RW drive
2:0,7,0: HBA     : (usb_msto,1) USB     USB HBA
   0,0,0: TAPE    : HP      C7438A          ZU5A <-- USB DAT Drive

In this example, this is a HP DAT72 Drive.

I have installed SCO OpenServer 6.0.0 or UnixWare 7.1.4, and was able to do a successful full system backup onto the USB DAT tape drive via the "emergency_rec ctape1" command.

I then created the emergency Recovery CD media via the command "emergency_disk cdrom1" and the creation was successful.

Please note that on UW714 + MP4 this option is hidden.

On the same hardware, I swapped the HD with a brand new empty one with the same size and booted the server with the Emergency Recovery CD, in order to do a system restore from the USB DAT Tape drive.

The boot via the Emergency Recovery CD was OK and got the expected message displayed:

The Hard Disk not sane
Then I get the menu list:
Access Openserver Shell
Restore Disk(s)
Reboot

When I choose "Restore Disk(s)" I get the error IN RED after it tels me To place the system backup tape:

"No tape has been detected in the Tape drive"....etc

I am then unable to perform a system restore from the USB DAT Tape drive, even though I was successful in performing a system backup to the same USB DAT Tape drive.

I had noticed when I went into the recovery shell via the "Access Openserver Shell" that the /dev/rmt directory was missing.

And when I went to the <Ctrl> <Alt> <Print Scr> <P> screen, it showed "sdiconfig -l" the USB DAT Tape drive to be listed.

I can return to the menu with:

<Ctrl> <Alt> <Print Scr> <F1>

HARDWARE: Any Server that comes with external/internal USB DAT Tape Drive

Cause

The Emergency Recovery Media creation under DRF has USB support but only a USB keyboard was taken into consideration when building the Emergency Recovery CD, hence this is why /dev/rmt/tape0 was not created, as the USB Tape was not taken into account.

Solution

This has been reported to SCO Engineering and is fixed in SCO OpenServer 6.0.0 MP3 & UnixWare 7.1.4 MP4.

Please note that the tape nodes (/dev/rmt/tape1) will only be created when you select the option to Restore from tape.

If you not have this Maintenance then as a workaround, please follow these steps as a fix:

1- Before creating the Emergency Recovery CD on the proto server,

cd to /usr/lib/drf and edit the drf_inst.gen file:

Go the section of the "get_sec_medium" function and add the following line just before the "for" loop over the /dev/rmt/tape

function get_sec_medium
{
   msg
   footer 2> /dev/null #clear the footer
   FOUND=false
   display "$TAPE_PROMPT"
   call getkey
   wclose
   # also create USB tape nodes      <--New added line
   /etc/scsi/bmkdev >/dev/null 2>&1  <--New added line
   for i in /dev/rmt/tape?
   do
       check_devs -g ${i} && {
               SEC_MEDIUM=$i
               FOUND=true
               break
       }
  done
2- Created the Emergency Recovery CD:
# emergency_disk cdrom1

3- Place a new HD (Same type as the proto one) and boot from the DRF CD, until you reach the expected menu list:

Access Openserver Shell
Restore Disk(s)
Reboot

4- Choose the "Restore Disk(s)" option and the DRF then will successfully detects The USB DAT tape and starts to restore.

5- Once the Restore is successful, you can now boot the restored OSR6 from the new HD.

NOTES:    TroubleShooting:
          ---------------
1- You will not be able to create an Emergency Floppy disk
   with OpenServer6.  The kernel size is too big for the floppy.

It will try though so it is recommended not to use:

# emergency_disk diskette1

2- You will, though, be able to write to a CD-ROM image using:

# emergency_disk cdrom1

3- If though, you do not have a writable CD-ROM then you will see the following error:

UX:drf: INFO: SONY CD-ROM CDU77E (2,0,0)

UX:drf: ERROR: cdrom1 seems to be a CD-ROM vs a CD-RW drive. Select a different cdrom or use the -i <image_dir> option.

4- To write an image, which can then be burned using a separate CD Burner, use the following:

# mkdir -p /local/cd.image
# emergency_disk -i /local/cd.image cdrom1

This will generate two files in /local/cd.image.

-rw-r--r--    1 root     sys      3128431 Jan 30 14:25 boot.cd.image
-rw-r--r--    1 root     sys      3493888 Jan 30 14:25 drf.image.iso
The first is the boot media.  boot.cd.image is a no-emulation
equivalent of the floppy image and as you can see it over 3MB in
size so it will not fit within 2.88 MB as this is the max size of a
floppy - hence only no-emulation mode CD boot is possible.

The file to burn, as directed by the "emergency_disk" command is:

drf.image.iso

5- To add additional files on the CD, then follow these steps:

# cd /usr/lib/drf
# cp drfram.proto drfram.proto.orig
Using your favourite editor, modify "drfram.proto":

To add a file use the following format:

Header -->
#ident  "@(#)
unixsrc:usr/src/i386/cmd/drf/cmd/drfram.proto
/main/legend/osr600mp2/2"

This is the root directory, it's type (d), permissions (--755) and it's owner (0 = root) and it's group (3 = sys) -->

d--775 0 3

Here are the files in the root directory, note that .sh_history is a link to /dev/null and TIMEZONE will sit in the root directory but will come from /etc/TIMEZONE on your server. -->

.sh_history     l--777 0 3 /dev/null
TIMEZONE        ---555 2 2 /etc/TIMEZONE

The files with $PROTO are reserved and some files like "step1rc" are required but may not be visible when the Emergency Boot CD boots and you access the shell (ie. the RAMDISK) -->

funcrc          ---755 2 2 $PROTO/stage/funcrc
inst            ---755 2 2 $PROTO/stage/drf_inst
step1rc         ---555 0 0 /usr/lib/drf/step1rc
step2rc         ---555 0 0 /usr/lib/drf/step2rc
Again, here we have the "isl" directory which sits under root and
contains the do_install file.  This comes from /inst on your server
and is linked to it -->
isl     d--755 0 3
        do_install      l--555 0 3 /inst
So, moving down, the official designation for additional files is to
add a "flop2" notation such as ones further down.  Here we are
adding a /usr/bin directory containing ls and dd from /u95/bin/ls -->
usr     d--775 0 3
        bin     d--755 0 3
#flop2          ls          ---555 0 3 /u95/bin/ls
#flop2          dd          ---555 0 3 /u95/bin/dd
Let's add in the "mapkey" binary:
#flop2          mapkey      ---555 0 3 /usr/bin/mapkey

To add your own, you can make up your own structures, for example to add in /etc/default/mapkey, go to etc then default and add: -->

default d--755 0 3
      workstations    ---444 0 3 $PROTO/work.1
#     fstyp   ---444 0 3 /etc/default/fstyp
      AFS     ---444 0 3 /etc/default/AFS
      EAFS    ---444 0 3 /etc/default/EAFS
      HTFS    ---444 0 3 /etc/default/HTFS
      S51K    ---444 0 3 /etc/default/S51K
      vxfs    ---444 0 3 /etc/default/vxfs
      fstyp   ---444 0 3 $PROTO/etc/default/fstyp
      mapkey  ---644 0 3 /etc/default/mapkey

You would also need to make a /usr/lib/keyboards directory and include the appropriate keyboard file from /usr/lib/keyboards.

NOTES:    As an additional workaround, for UnixWare 7.1.4 MP4 please follow
          these steps as a fix:

1- Before creating the Emergency Recovery CD on the proto server, note the major/minor number of the /dev/rmt/ctape1 device node. Im our case it was 296/101:

# l /dev/rmt/ctape1
crw-rw-rw-  2 root   sys    296, 101 Jan  9 12:26 /dev/rmt/ctape1

2- cd to /etc/lib/drf and edit the drfram.proto file and add these entries:

rmt     d--775 0 3
        tape1   c--666 0 3 296 51
        $
after the line
zero c--666 0 3 39 1
$

The minor number above is from (Original Minor # - 50). (101 - 50) = 51.

You need to subtract 50 from the original Minor number for the workaround to work.

This issue has been reported to SCO Engineering.

From the Xinuos technical article database, record 126683. Reproduced with permission. The original is at www.sco.com/ta/126683, which is reachable but blocked from search engines.

Next step

Need this fixed rather than explained?

Tell us the release, what the system runs and what it is doing. We will reply by email with what we would do next.

  • You deal with the people who do the work
  • No obligation, and no upgrade required before we will help
  • We tell you plainly if something is not worth doing

Send an enquiry

We reply by email. If a server is down, say so in the first line. You can also write to help@scoexpert.com.