cpio screw up - SGI

This is a discussion on cpio screw up - SGI ; Hi, I'm trying to recover from a disaster on 2 of my 120Gb hard drives. I've got backups on DLT4000 tapes ;-) But... when I try to recover I get the following error message Bad magic number/header. 1 error any ...

+ Reply to Thread
Page 1 of 2 1 2 LastLast
Results 1 to 20 of 21

Thread: cpio screw up

  1. cpio screw up

    Hi,
    I'm trying to recover from a disaster on 2 of my 120Gb hard drives. I've
    got backups on DLT4000 tapes ;-) But... when I try to recover I get the
    following error message

    Bad magic number/header. 1 error

    any ideas? Also I've tried dd if=/dev/tape of=/usr/data/1.cpio bs=2048k
    then playing with the file. Is there anyway I can bin the first say 10mb's
    of the file and recover the rest? The data I'm recovering is quite
    important and I'm having probs with both sets of backup tapes..... Not
    good...

    The other tapes give the following error message

    Corrupt header, file(s) may be lost.

    (340 blocks) Re-synchronized on magic number/header.
    0 blocks
    681 error(s)
    Errors during restore.



    any idea how I can recover the data from the tapes. I'm 100% sure it is
    possible just no idea how. Using the -k option on cpio didn't seem to do
    much to help either :-(

    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH
    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


  2. Re: cpio screw up

    In article ,
    Khalid Schofield wrote:
    :I'm trying to recover from a disaster on 2 of my 120Gb hard drives. I've
    :got backups on DLT4000 tapes ;-) But... when I try to recover I get the
    :following error message

    :Bad magic number/header. 1 error

    :any ideas? Also I've tried dd if=/dev/tape of=/usr/data/1.cpio bs=2048k
    :then playing with the file. Is there anyway I can bin the first say 10mb's
    f the file and recover the rest?

    Try searching for the magic number 070707 (octal), but not necessarily
    on a block boundary. See man 4 cpio for the header structure.
    --
    "The human genome is powerless in the face of chocolate."
    -- Dr. Adam Drewnowski

  3. Re: cpio screw up

    * Khalid Schofield
    | I'm trying to recover from a disaster on 2 of my 120Gb hard
    | drives. I've got backups on DLT4000 tapes ;-) But... when I try to
    | recover I get the following error message
    |
    | Bad magic number/header. 1 error
    |
    | any ideas?

    It might help if you described how
    - the backup tapes were created
    - what you tried to restore the backups
    (both: show the commands you typed)

    R'

  4. Re: cpio screw up

    Hi,
    Well I used the irix backup gui that uses cpio. I'm using the regular irix
    cpio and not gnu-cpio. I've backed up all of my data on a 120Gb drive. The
    drive was xfs format (doesn't matter though) and I'm restoring to the same
    drive after fx -x'ing it and re-formating it as xfs. Trying to restore
    video files, audio, all my emails, notes etc etc. Pretty much all my
    useful data on the box. D'oh....disk corruptions suck,

    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH

    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


    On Fri, 7 Nov 2003, Ralf Fassel wrote:

    > * Khalid Schofield
    > | I'm trying to recover from a disaster on 2 of my 120Gb hard
    > | drives. I've got backups on DLT4000 tapes ;-) But... when I try to
    > | recover I get the following error message
    > |
    > | Bad magic number/header. 1 error
    > |
    > | any ideas?
    >
    > It might help if you described how
    > - the backup tapes were created
    > - what you tried to restore the backups
    > (both: show the commands you typed)
    >
    > R'
    >


  5. Re: cpio screw up

    * Khalid Schofield
    | Well I used the irix backup gui that uses cpio.

    Do you know how they write the header (check -c option and related for
    cpio)? Sounds like you might try to read a binary header as ASCII or
    vice versa. Other than that I would try to check
    /usr/sbin/Backup|Restore for details how these create the tapes...

    Good luck...
    R'

  6. WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    Hi,
    well I'm looking at that now. I guess what the route of the problem is
    that I'm using the ACARD IDE-SCSI bridges. I really should have my data on
    proper SCSI drives. I've been looking for some 72Gb or larger scsi drives
    on ebay but havn't found any that will work in Octane's internal bus. Do
    you or any one elce reading this know where I can get big scsi drives for
    under 200 pounds?

    I've been looking at
    http://cgi.ebay.co.uk/ws/eBayISAPI.d...hippingPayment

    Will this work in an Octane?

    Well when I've got the tapes to restore I don't want this to happen again
    so I guess I'll go get SCSI drives like others have advised me to. Just
    need to find some and at the right price.

    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH

    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


    On Fri, 7 Nov 2003, Ralf Fassel wrote:

    > * Khalid Schofield
    > | Well I used the irix backup gui that uses cpio.
    >
    > Do you know how they write the header (check -c option and related for
    > cpio)? Sounds like you might try to read a binary header as ASCII or
    > vice versa. Other than that I would try to check
    > /usr/sbin/Backup|Restore for details how these create the tapes...
    >
    > Good luck...
    > R'
    >


  7. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    On Fri, 7 Nov 2003 20:01:54 +0000, Khalid Schofield
    wrote:

    >Hi,
    >well I'm looking at that now. I guess what the route of the problem is
    >that I'm using the ACARD IDE-SCSI bridges. I really should have my data on
    >proper SCSI drives. I've been looking for some 72Gb or larger scsi drives
    >on ebay but havn't found any that will work in Octane's internal bus. Do
    >you or any one elce reading this know where I can get big scsi drives for
    >under 200 pounds?


    You can buy a new Maxtor 73GB U160 10k SCA for 161 from www.scan.co.uk :-)

    Cheers

    Neil

    --
    -------------------------------------------------
    Neil Rothwell
    Freelance Mechanical Design Engineer
    E-Mail: neil@rothers.demon.co.uk
    Home Page: http://www.rothers.demon.co.uk
    -------------------------------------------------

  8. Re: cpio screw up

    Dans article ,
    oums0246@raven.linux.ox.ac.uk disait...
    >
    > Hi,
    > Well I used the irix backup gui that uses cpio.


    It used to be bru, too (depends upon the IRIX version)...

    --
    Quis, quid, ubi, quibus auxiliis, cur, quomodo, quando?

  9. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    YAY!
    just off to order one
    thanks ;-)

    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH

    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


    On Fri, 7 Nov 2003, Neil Rothwell wrote:

    > On Fri, 7 Nov 2003 20:01:54 +0000, Khalid Schofield
    > wrote:
    >
    > >Hi,
    > >well I'm looking at that now. I guess what the route of the problem is
    > >that I'm using the ACARD IDE-SCSI bridges. I really should have my data on
    > >proper SCSI drives. I've been looking for some 72Gb or larger scsi drives
    > >on ebay but havn't found any that will work in Octane's internal bus. Do
    > >you or any one elce reading this know where I can get big scsi drives for
    > >under 200 pounds?

    >
    > You can buy a new Maxtor 73GB U160 10k SCA for 161 from www.scan.co.uk:-)
    >
    > Cheers
    >
    > Neil
    >
    > --
    > -------------------------------------------------
    > Neil Rothwell
    > Freelance Mechanical Design Engineer
    > E-Mail: neil@rothers.demon.co.uk
    > Home Page: http://www.rothers.demon.co.uk
    > -------------------------------------------------
    >


  10. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    Benjamin Goldsteen wrote:
    >
    > How good are Maxtor's SCSI drives? In the past, I've always had the
    > best experience with the Seagate. The worst with Quantum.


    Didn't Maxtor buy Quantum (or at least their SCSI division)?

  11. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    If I remember rightly yup they did. Quantum if I remember rightly are the
    peeps who produce the DLT tape drives. I ordered an U160 SCA 80 10,000 rpm
    72Gb drive yesterday. Can't wait to get it in my desktop (my octane). Are
    there any jumpers to set on the drive to tell it to use Single Ended and
    not LDV?

    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH

    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


    On Sat, 8 Nov 2003, John-Paul Stewart wrote:

    > Benjamin Goldsteen wrote:
    > >
    > > How good are Maxtor's SCSI drives? In the past, I've always had the
    > > best experience with the Seagate. The worst with Quantum.

    >
    > Didn't Maxtor buy Quantum (or at least their SCSI division)?
    >


  12. Re: cpio screw up

    I've managed to recover from the old tapes but still can't get the most
    recent tapes to recover. I've dd off a 1Gb sample to play with off of tape
    1. It looks like the tapes index / catalog is corrupt. The binary data
    must still be there that contains the files internal data. If I dd off an
    intire tape how can I throw away the beginning of the file and use cpio to
    recover the rest? This must be possible using some technique hell it's
    unix anything is possible ;-)

    Well I'll have another play today. Hope I get a little further. If anyone
    coes up with any bright ideas then please let me know.


    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH

    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


    On Fri, 7 Nov 2003, Emmanuel Florac wrote:

    > Dans article ,
    > oums0246@raven.linux.ox.ac.uk disait...
    > >
    > > Hi,
    > > Well I used the irix backup gui that uses cpio.

    >
    > It used to be bru, too (depends upon the IRIX version)...
    >
    > --
    > Quis, quid, ubi, quibus auxiliis, cur, quomodo, quando?
    >


  13. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up


    > Hi,
    > well my server has been going just over a year. Chugging away with it's
    > 1Tb raid. I've used ide drived at home in my server. Very sad experience.
    > Had 2 drives fail..... The ole server at work hasn't been going long but
    > we did have a 10 year old hp running with maxtor drives if I remember
    > rightly in a raid0 config removed from service this year :-). Just how are
    > the scsi drives different in construction to ide drives? Why are they so
    > much better?


    Because MTBF is not as important to IDE manufacturers. IDE customers tend
    to be home users, who value cost and density more than reliability.



  14. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    "Paul" underscores@no.spaa-aam.yahoo.com> writes:
    > > well my server has been going just over a year. Chugging away with it's
    > > 1Tb raid. I've used ide drived at home in my server. Very sad experience.
    > > Had 2 drives fail..... The ole server at work hasn't been going long but
    > > we did have a 10 year old hp running with maxtor drives if I remember
    > > rightly in a raid0 config removed from service this year :-). Just how are
    > > the scsi drives different in construction to ide drives? Why are they so
    > > much better?

    >
    > Because MTBF is not as important to IDE manufacturers. IDE customers tend
    > to be home users, who value cost and density more than reliability.


    Yes, but the ways to handle a finite MTBF are regular backups and RAID
    setups. Because SCSI drives will fail just as surely as IDE drives,
    running a raid 0 on either of them with something else than
    temporary/inexpensive data in the corresponding filesystems is asking
    for trouble anyway.

    And before I forget: most IDE drives are optimized for regular
    downtime. IDE drives which are not specified for continuous operation
    (which includes quite a share of todays market) often have some fluid
    that can only be returned to certain places in the drive when a
    spin-down occurs.

    Thomas Jahns
    --
    "Computers are good at following instructions,
    but not at reading your mind."
    D. E. Knuth, The TeXbook, Addison-Wesley 1984, 1986, 1996, p. 9

  15. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up


    > > Because MTBF is not as important to IDE manufacturers. IDE customers

    tend
    > > to be home users, who value cost and density more than reliability.

    >
    > Yes, but the ways to handle a finite MTBF are regular backups and RAID
    > setups. Because SCSI drives will fail just as surely as IDE drives,
    > running a raid 0 on either of them with something else than
    > temporary/inexpensive data in the corresponding filesystems is asking
    > for trouble anyway.


    Agreed, and not with just disk media. Only having one copy of data that has
    any value is not smart. That goes for any RAID configuration, tape media,
    optical media, etc. But the post asked for reasons why IDE drives fail more
    frequently than SCSI drives, and that reason is that they are not engineered
    for availability.



  16. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    oN 09-Nov-03, Paul said:

    > Because MTBF is not as important to IDE manufacturers. IDE customers
    > tend to be home users, who value cost and density more than
    > reliability.


    Partially true. The new SATA drives from Seagate, WD, and Maxtor, at
    120GB and above, at least, are warranted for 3 years.

    --
    Bill
    Posted with XanaNews Version 1.15.7.4

  17. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    On Sun, 9 Nov 2003 08:15:49 +0000, Khalid Schofield
    wrote:

    >If I remember rightly yup they did. Quantum if I remember rightly are the
    >peeps who produce the DLT tape drives.


    Quantum was spun-off from DEC before DEC disappeared.

    >I ordered an U160 SCA 80 10,000 rpm
    >72Gb drive yesterday. Can't wait to get it in my desktop (my octane). Are
    >there any jumpers to set on the drive to tell it to use Single Ended and
    >not LDV?


    (LVD, not LDV) No, that's done by a wire in the interface (/DIFFSENS).

    And the whole point of SCA and SCA-2 is to remove the need for
    on-drive jumpers....

    >Khalid Schofield
    >System Administrator / EM Technician
    >Dept. Of Materials


    Malc.

  18. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    Thomas Jahns writes:

    > And before I forget: most IDE drives are optimized for regular
    > downtime. IDE drives which are not specified for continuous operation
    > (which includes quite a share of todays market) often have some fluid
    > that can only be returned to certain places in the drive when a
    > spin-down occurs.


    I'd be most interested in a reference for that information.

    *p


  19. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    Great ;-)
    glad to hear I just chuck the disk in :-) I do obsessivly back everything
    up to my DLT4000 tape's on a regulat basis but it is nice to know that the
    drive is far more robust than the ones that I'm having probs with at the
    moment.

    *********************
    Khalid Schofield
    System Administrator / EM Technician
    Dept. Of Materials
    University Of Oxford
    Parks Road
    Oxford
    OX1 3PH

    Email: khalid.schofield@materials.ox.ac.uk
    Tel: 01865 273785
    Fax: 01865 283333
    Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


    On Mon, 10 Nov 2003, Malcolm Weir wrote:

    > On Sun, 9 Nov 2003 08:15:49 +0000, Khalid Schofield
    > wrote:
    >
    > >If I remember rightly yup they did. Quantum if I remember rightly are the
    > >peeps who produce the DLT tape drives.

    >
    > Quantum was spun-off from DEC before DEC disappeared.
    >
    > >I ordered an U160 SCA 80 10,000 rpm
    > >72Gb drive yesterday. Can't wait to get it in my desktop (my octane). Are
    > >there any jumpers to set on the drive to tell it to use Single Ended and
    > >not LDV?

    >
    > (LVD, not LDV) No, that's done by a wire in the interface (/DIFFSENS).
    >
    > And the whole point of SCA and SCA-2 is to remove the need for
    > on-drive jumpers....
    >
    > >Khalid Schofield
    > >System Administrator / EM Technician
    > >Dept. Of Materials

    >
    > Malc.
    >


  20. Re: WANTED BIG SCSI DRIVES- route of problem Re: cpio screw up

    Per Ekman writes:
    > Thomas Jahns writes:
    >
    > > And before I forget: most IDE drives are optimized for regular
    > > downtime. IDE drives which are not specified for continuous operation
    > > (which includes quite a share of todays market) often have some fluid
    > > that can only be returned to certain places in the drive when a
    > > spin-down occurs.

    >
    > I'd be most interested in a reference for that information.


    This is from German c't magazine issue 19/1998 page 144.

    Please forgive my amateurish translation:

    ...while Fujitsu, Maxtor and Seagate did not disapprove continuous
    operation of their desktop models, an IBM employee gave us the
    surprising hint that IBM's desktop drives would live longer if spun
    down from time to time. The reason for this lies in the different
    coating of disks in desktop versus server drives which is applied to
    improve on the start/stop cycle tolerance of the drive. From this
    coating comes the disadvantage that in continuous use the lubricant
    will cover the read/write heads. This problem is prevented by turning
    off the drive because the heads will then be cleaned in the specially
    textured landing zone.

    The article also mentions that even entry-level server drives (which
    are SCSI) have a user adjustable ADM function (automatic drive
    maintenance) which would automatically take spin down the drive after
    one week of continuous operation. In the DCAS and DORS series the
    function was turned on by default, in the DDRS series it was turned
    off.

    All this was presented in a discussion about how many on/off cycles
    different drives could handle (does switching off of hard disk drives
    damage the drive apart from saving power and reducing noise?).

    As the information is quite dated I should correct my memory:

    There once was a time when desktop drives of one manufacturer required
    downtime in regular intervals. (Although that one manufacturer then
    indeed represented a very big part of the local market.)

    Wether this applies equally to todays IDE drives I cannot be sure.

    It's fascinating how fuzzy my memory becomes after just five years.

    Thomas Jahns
    --
    "Computers are good at following instructions,
    but not at reading your mind."
    D. E. Knuth, The TeXbook, Addison-Wesley 1984, 1986, 1996, p. 9

+ Reply to Thread
Page 1 of 2 1 2 LastLast