[OpenAFS] Cleaning up the backup-db

[Chris Huebsch]

Hello,

I am using the integrated afs-backup for a file-based backup-system.

I have 3 AFS-Servers with each 400 GB of diskspace.
(afs1, afs2, and afs3)

I have the following vol-sets:

|Volume set user1:
|    Entry   1: server afs1, partition vicepa, volumes: user\..*\.backup
|
|Volume set user2:
|    Entry   1: server afs2, partition vicepa, volumes: user\..*\.backup
|
|Volume set user3:
|    Entry   1: server afs3, partition vicepa, volumes: user\..*\.backup
|
|Volume set media:
|    Entry   1: server .*, partition .*, volumes: media\..*
|    Entry   2: server .*, partition .*, volumes: media
|
|Volume set www:
|    Entry   1: server .*, partition .*, volumes: www\..*
|    Entry   2: server .*, partition .*, volumes: www
|
|Volume set admin:
|    Entry   1: server .*, partition .*, volumes: root\..*
|    Entry   2: server .*, partition .*, volumes: home
|    Entry   3: server .*, partition .*, volumes: admin\..*
|    Entry   4: server .*, partition .*, volumes: admin
|
|Volume set ftp:
|    Entry   1: server .*, partition .*, volumes: ftp\..*
|    Entry   2: server .*, partition .*, volumes: ftp

(I know, there is a little bit redundancy in it...)

I have the following dumpsets:

|/Sunday2  expires in  21d
|    /Monday2  expires in  14d
|    /Tuesday2  expires in  14d
|    /Wednesday2  expires in  14d
|    /Thursday2  expires in  14d
|    /Friday2  expires in  14d
|    /Saturday2  expires in  14d
|/Sunday1  expires in  21d
|    /Monday1  expires in  14d
|    /Tuesday1  expires in  14d
|    /Wednesday1  expires in  14d
|    /Thursday1  expires in  14d
|    /Friday1  expires in  14d
|    /Saturday1  expires in  14d
|/Sunday0  expires in  21d
|    /Monday0  expires in  14d
|    /Tuesday0  expires in  14d
|    /Wednesday0  expires in  14d
|    /Thursday0  expires in  14d
|    /Friday0  expires in  14d
|    /Saturday0  expires in  14d

I do a file-based backup, with all 3 AFS-Server serving as bupc-Server.

The backup is initiated by a cron-job:

|WEEKNUM=$(($(date +"%U") % 3)) #find the number of the week mod 3
|DAY=$(${DATE} +"%A")           #name of the day
|
|JOBFILE="/tmp/$$.afsbackup"
|if [ $DAY = "Sunday" ]         #full backup on Sunday
|then
|  #       volset dumplevel dumptarget
|  cat <<EOF > $JOBFILE
|  dump user1 /${DAY}${WEEKNUM} 2
|  dump user2 /${DAY}${WEEKNUM} 3
|  dump user3 /${DAY}${WEEKNUM} 1
|  dump media /${DAY}${WEEKNUM} 3
|  dump www   /${DAY}${WEEKNUM} 3
|  dump admin /${DAY}${WEEKNUM} 2
|  dump ftp   /${DAY}${WEEKNUM} 2
|EOF
|else                           #incremental backup on weekday
|  #       volset dumplevel dumptarget
|  cat <<EOF > $JOBFILE
|  dump user1 /Sunday${WEEKNUM}/${DAY}${WEEKNUM} 2
|  dump user2 /Sunday${WEEKNUM}/${DAY}${WEEKNUM} 3
|  dump user3 /Sunday${WEEKNUM}/${DAY}${WEEKNUM} 1
|  dump admin /Sunday${WEEKNUM}/${DAY}${WEEKNUM} 2
|EOF
|fi
|
|$BACKUP -file $JOBFILE 2>/dev/null >/dev/null
|backup savedb 1 -localauth 2>/dev/null >/dev/null
|backup savedb 2 -localauth 2>/dev/null >/dev/null
|backup savedb 3 -localauth 2>/dev/null >/dev/null

Now the dump-records start to pile up.

My dumps do have an expiry-time. According to "Removing Obsolete Records from
the Backup Database", old records are deleted, when the tape is
recycled. But I do not have tapes!

So I tried 'backup deletedump -to "..."'. The ordinary dump-records can be
deleted this way. But the Ubik_db_dump-records stay in the db:

1115452908          0 0  05/07/2005 10:01  1     0 Ubik_db_dump

I have to delete them with:

'backup deletedump -id "$id"'

That is very annoing, because it will complicate my scripts a lot.
Is there a better way doing this?

Is it a bug or a feature?



Chris
-- 
Chris Huebsch    www.huebsch-gemacht.de | TU Chemmnitz, Informatik, RNVS
GPG-Encrypted mail welcome! ID:7F2B4DBA |   Str. d. Nationen 62, B204
  Chemnitzer Linux-Tage 2006, 4.-5.Maerz |       D-09107 Chemnitz
     http://chemnitzer.linux-tage.de/    |  +49 371 531-1377, Fax -1803