Read the Backup Retention Policy.
View backup logs by date: Backup Archive
Beginning with the November 2004 backup, the volume set and dump schedule names have been changed to be a bit more meaningful. The volume sets are defined as follows:
backup> listvolsets Volume set all: Entry 1: server babylon.citi.umich.edu, partition .*, volumes: .*\.backup backup>Which means that volumeset all includes the volumes on server babylon.citi.umich.edu, on all of its partitions (.*), which are named xxx.backup (.*\.backup).
The dump schedule is defined as follows:
backup> listdumps /full /wek1 /daily /wek2 /daily /wek3 /daily /daily backup>This schedule assumes that a full dump (/full) will be performed once a month. Weekly incremental dumps may be done in the intervening weeks. (/full/wek1, /full/wek2, /full/wek3) Daily dumps may be performed off the monthly full dump until a weekly is performed (/full/daily), or off of a previous weekly dump (/full/wek2/daily).
This has the effect of changing the name of the tapes. A full dump of the volume set all will have tapes named like 'all.full.1', 'all.full.2'. A weekly incremental may be named 'all.wek3.1'.
OpenAFS Backup Instructions
Here are the typical backup commands that might be used to do a backup:
The screamer backups are automated with scripts. /usr/local/bin/dump-to-got-back.sh is run out of root's cron and does daily incrementals. It assumes the incremental tape is in the drive, goes to the current end of data on the tape, and writes the new incremental. The script sends e-mail upon successful or unsuccessful completion.
When the incremental tape is filled, the automated dump will fail, and a new full (level-0) dump is needed. This is done manually by putting in a new tape to hold the full dump and running /usr/local/bin/dump-to-back-full.sh .
Backups are run by kwc