faq

一套客户/服务器模式的备份系统代码,跨平台,支持linux,AIX, IRIX, FreeBSD, Digital Unix (OSF1), Solaris and HP-UX.

                AF's Backup FAQ
                ===============

Index
-----

Q1: How do I tell how much space is left on the tape?

Q2: Why is the mtime used for deciding what files to save during incremental
    backup and not the ctime or both ?

Q3: Do my current configurations get overwritten during an upgrade ?

Q4: Why should I and how do I use sets of cartridges ?

Q5: How many cartridges should I use ?

Q6: I have a robot with n cartridges. Can i use more than n tapes ?

Q7: Can ordinary users restore their own files and directories ?

Q8: Why does afbackup not have a GUI ?

Q9: What does the warning mean: "Filelist without user-ID information ..." ?

Q10: The whole backup systems hangs in the middle of a backup, what's up ?

Q11: Tape reels back and forth, mail sent "tape not ready ...", what's up ?

Q12: The server seems to have a wrong tape file count, what's wrong ?

Q13: When using crond, the client seems not to start correctly ... ?

Q14: What does AF mean ?

Q15: Though client backup works, remote start does not. Why ?

Q16: My server does not work, tape operates, but nothing seems to be written ?

Q17: I have a ADIC 1200G autoloading DAT and no docs. Can i use it with HPUX ?

Q18: What is a storage unit and how and why should i use it ?

Q19: Why should i limit the number of bytes per tape ?

Q20: What are backup levels and why should i use them ?

Q21: What do all the files in the var-directories mean ?

Q22: Help ! My (multi stream server's) client receives no data, why ?

Q23: My DLT writes terribly slowly and changes cartridge too early, why ?

Q24: When should i use the multi stream server and when not ?

Q25: Why is my 2 GB capacity DAT tape full having written about 1.3 GB ?

Q26: Tape handling seems not to work at all, what's wrong ?

Q27: How can i change the compression configuration ?

Q28: Why does my Linux kernel Oops during afbackup operation ?

Q29: Why does afbackup not use tar as packing format ?

Q30: How to recover directly from tape without client/server ?

Q31: Why do files truncated to multiples of 8192 during restore ?

Q32: What is the difference between total and real compression factor ?

Q33: How does afbackup compare to amanda ?

Q34: How to contribute to I18N/L10N ?

Q35: Why does I18N not work in my environment ?

Q36: Is there a mailing list or a home page for afbackup ?

Q37: I have trouble using the multi stream server. What can i do ?

Q38: On AIX i get the warning: decimal constant is so large ... what's that ?

Q39: What about security ? How does this authentication stuff work ?

Q40: Why does remote start of backups not work, while local start does ?

Q41: What is the architecture of afbackup ?

Q42: Why are new files with an old timestamp not saved during incr_backup ?

Q43: What do the fields in the minimum restore info mean ?

Q44: What are those files like /tmp/afbsp_XXXXXXX ? Can i remove them ?

Q45: On a client starting remotely i see a warning about no start cartridge
     found in the server log. What does it mean and can i suppress that ?



Questions and Answers
=====================

Q1: How do I tell how much space is left on the tape?

A1: This is hard to tell due to the problem to determine exactly, how many
    bytes can be written to a certain physical tape. Since version 3.1 the
    server counts the bytes written to each tape. The sums are written into
    the file /path/to/server/var/bytes_on_tape , one entry per line in the
    format cartridge-number colon number-of-bytes-on-tape . If a tape is
    full and the next one is automatically inserted, the number tells you
    the number of bytes, the server was able to write to tape, but here the
    streamer device might have applied compression, so the real number of
    bytes on tape may be smaller. If clientside compression is turned on,
    it is quite unlikely, that the streamer was able to even further pack
    the data, so in this case the number logged to the file should be close
    to reality.
     One method to find out the tape capacity is as follows:
    - tape a gzip-ped file (with -9), that is larger than 1 MB compressed
    - put an empty and unused tape into the streamer device, all data on
      it will be lost during this test
    - in a csh run the command:
        repeat 100000 cat filename | dd of=/dev/st0 obs=1048576
      (replace filename and /dev/st0 appropriately with the name of the
       gzip-ped file and the real streamer device)
    - the command will write the tape until end of media is reached and
        will output something like:
           4036+1 records in
           4036+1 records out
        This tells you, that 4036 * 1048576 i.e. 4036 MB were successfully
        written to tape.
    It must be kept in mind, that this test does not consider the space
    between tape files. Each time, a new tape file starts, a file space
    is written to tape, that wastes tape capacity of about 2 MB each. Refer
    to the documentation of your streamer device.


Q2: Why is the mtime used for deciding what files to save during incremental
    backup and not the ctime ?

A2: First: The ctime changes any time a chmod, a chown, or other operations
      modifying the inode are performed. A change like this is not worth
      selecting this file for backup, cause the file itself did not change.
      BTW the ctime can be evaluated additionally to the mtime setting the
      client side parameter UseCTime . But then the access time (atime) is
      not restored to the previous value after backup.
    Second: After backup of a file, afbackup restores the atime, because
      i found the atime a quite worth information. A restore of the atime
      changes the ctime, no way around this. If the ctime was evaluated
      for choosing the files for incremental backup, a file stored once
      would be saved again all the following backups, cause at any
      backup time the ctime changes. Incremental backup would be senseless,
      cause all files would be saved all the time a backup runs.


Q3: Do my current configurations get overwritten during an upgrade ?

A3: No. Nothing gets overwritten or lost. Newly introduced parameters have
    the old non-configurable behaviour as default. The defaults are applied,
    if the appropriate parameters are not given explicitly in the
    configuration files.


Q4: Why should I and how do I use sets of cartridges ?

A4: The question, why, is not that easy to answer. Maybe you have groups
      of hosts you would like to save to distinguished cartridges, maybe
      you would like to make the full backups to other cartridges than the
      incremental backup. Maybe you have the requirement, that you want to
      use an infinite number of cartridges for the full backup and reuse
      the ones for the incremental backup each time another full backup
      has finished. Or you might want to restrict the access to sets of
      cartridges to certain machines. Then you can configure access lists.
      Maybe you have more exotic requirements ...
    The answer to the how is easy: Set the serverside parameter
      Cartridge-Sets. The specifiers for the sets must be separated by
      whitespace and each may consist of digits, commas and dashes, e.g.
      3,6-9 . A set may be a single cartridge, but i do not recommend
      this, cause writing to the beginning of that cartridge destroys the
      rest stored on it and making these data unaccessible. The last
      number is usually the number of cartridges you have, but not
      necessarily. Cartridges at the upper end of the numbers might be
      omitted. If the last number is not equal to the number of cartridges,
      this number is NOT automatically added. The many numbers are given
      with this parameter, the many cartridge sets you have. The default,
      if this parameter is not present, is one cartridge set with all
      available cartridges. Enter man afserver.conf for details, how to
      configure client access restrictions for each set individually.


Q5: How many cartridges should I use ?

A5: The cartridges should be have enough capacity for at least two times
    a full backup including subsequent incremental backups. Otherwise
    files could get lost due to an unsuccessful backup overwriting
    previously stored data.


Q6: I have a robot with n cartridges. Can i use more than n tapes ?

A6: This question is obsolete as of afbackup version 3.3. This version
    maintains a cartridge database and allows to configure commands
    for media changers. Cartridge numbers and slot numbers need not to
    have anything to do with each other. See the HOWTO Q26.

    This is obsoleted text:
    Yes, you can use any number of tapes, if your robot is in the
    sequential mode. Simply fake a higher number to the backup system
    in the serverside configuration file. The only point is, that you
    have to change the cartridges in the robot manually in time. If
    you have e.g. a robot with 10 cartridges and would like to use 20,
    then you have to watch, when it is time to insert other cartridges
    to the appropriate positions. E.g. when cartridge number 8 is in
    the drive, take out cartridges 1-7 and insert number 10-16 into
    the appropriate slot. Later, when they are in use, you can replace
    8-10 by 17-19 and so on.
     When you want to do a restore, the restore-program tells you,
    where it wants to read from like this:

    Going to restore from cartridge X, file Y ...

    Insert manually the right cartridges into slots, the robot will
    access next time. The system will automatically recognize by the
    label on the tape, that it has found the right cartridge now. A
    warning is written to the serverside log telling, that another
    cartridge was found than expected, but this is just a warning and
    we know, how this happened ...

    The patterns %b, %c, %m and %n might be helpful in the server's
    Change-Cart-Command. They are replaced as follows:

     %c   The number of the cartridge currently in the drive
     %b   The number of the cartridge currently in the drive minus 1
     %n   The number of the cartridge expected to be put into the
          drive after ejecting. The cartridge handler must be in
          sequential mode. If no cartridge handler is present, %n
          will not be replaced.
     %m   like %n, but 1 is subtracted

    So e.g. if you have groups of 10 cartridges each to be put into
    the cartridge handler and want to be informed each time the 10th,
    20th, ... cartridge is ejected and you might want to change the
    cartridges, you can write a small script as a wrapper for the mt
    command. Let's call this script eject_check_10 with the device,
    the current cartridge number, the expected next cartridge number
    and a user to be E-mailed as arguments. Configure this command
    as the Change-cart-command like this:
     /your/path/to/eject_check_10 %d %c %n Backupmaster
    The script itself might look like this:

#!/bin/sh
#
# Usage: eject_check_10 <device> <current-cartno> <next-cartno> <mailaddr>
#

DEVICE="$1"
CURRENTCART="$2"
NEXTCART="$3"
MAILADDR="$4"

TMPFILE=/tmp/cart_group_markerfile.$$
/bin/rm -f $TMPFILE

CURRENTGROUP=`expr '(' $CURRENTCART - 1 ')' / 10`
NEXTGROUP=`expr '(' $NEXTCART - 1 ')' / 10`

if [ $CURRENTGROUP -ne $NEXTGROUP ] ; then
  touch $TMPFILE

  mail "$MAILADDR" << END_OF_MAIL
    Hello,

    Please insert new cartridges into your cartridge handler.
    The current cartridge is $CURRENTCART and the expected next
    one is $NEXTCART. Remove the file $TMPFILE, when done.

    Regards, your automatic backup service

END_OF_MAIL

  while [ -f $TMPFILE ] ; do
    sleep 5
  done

else

  # simply perform the eject
  #
  exec mt -f $DEVICE rewoffl

fi