backup-system.conf.example

# /etc/backup-system.conf — per-host settings for linux-backup-system.
#
# deploy.sh generates this from detection; it is safe to hand-edit. Every value
# has a built-in default in backup-common.sh, so a missing line is fine. Nothing
# about the boot layout goes here — that is derived at run time.

# Where the backup drive is mounted, and the borg repo inside it.
BACKUP_MOUNT="/mnt/backup"
#BORG_REPO="/mnt/backup/borg-backup"      # default: $BACKUP_MOUNT/borg-backup

# Wrong-drive guard: the backup filesystem's UUID. The backup scripts refuse to
# run if the drive mounted here is not this one. Leave empty to disable.
BACKUP_FS_UUID=""

# Unlock-on-attach (external drives only): the drive's LUKS partition UUID and a
# keyfile enrolled in one of its keyslots. Used by borg-backup-drive-attach.sh.
BACKUP_LUKS_UUID=""
BACKUP_KEYFILE="/etc/luks-keys/backup-drive.key"
#BACKUP_MOUNT_OPTS="compress=zstd:1"      # btrfs example; leave empty on ext4/xfs

# Scheduling policy:
#   adhoc     — external/removable drive; backups run by hand or on drive-attach,
#               timers are NOT enabled. (default; correct for most boxes)
#   scheduled — second INSTALLED drive; the daily borg/BIT timers are enabled.
SCHEDULE_MODE="adhoc"

# Extra absolute mountpoints to include in file-level backups, space-separated.
# Every separately mounted local filesystem (/, /home, /boot, the ESP, an
# openSUSE /var, a ZFS dataset…) is found automatically; the backup drive,
# removable media, network shares and bind mounts are left out.
BACKUP_EXTRA_SOURCES=""

# Retention — applied by every layer (borg, btrfs/Timeshift replicas, BIT).
# Count-based plus free-space based, NEVER time-based: an ad-hoc drive may sit
# unplugged for months and age must never delete a backup.
KEEP=10             # normal: keep the newest KEEP of each set, any age
MIN_KEEP=3          # hard floor: never prune below this, even when the drive is full
MIN_FREE_PCT=10     # keep at least this % of the drive free
MIN_FREE_GIB=0      # and at least this many GiB free (0 = ignore)

# Capacity — checked before a drive is formatted or adopted, and before every
# backup. The drive must hold one full copy of everything in the backup sources
# plus CAPACITY_HEADROOM_PCT spare room (measured against actual usage) or it
# is refused; it should be CAPACITY_RECOMMEND_X times the total size of the
# Linux filesystems it backs up (not the whole disk — a dual-boot or Apple
# Silicon disk is mostly another OS), or the suite says so (a warning, not a stop).
CAPACITY_HEADROOM_PCT=20
CAPACITY_RECOMMEND_X=2

# The name this host's backups are filed under: the borg archive prefix, the
# Back In Time chain directory, the replica names. deploy.sh pins the hostname
# at install time so a later hostname change does not orphan the chain.
#BACKUP_HOST_ID="myhost"

# Extra exclude patterns for the file-level layers (borg and Back In Time
# share one list, see bx_excludes in backup-common.sh), space-separated,
# anchored at /: BACKUP_EXTRA_EXCLUDES="/home/*/VMs/* /srv/scratch/*"
BACKUP_EXTRA_EXCLUDES=""

# Paths re-included INSIDE an excluded tree (they win over every exclude),
# space-separated. Keep configuration and keys, drop the bulk data:
#   BACKUP_EXTRA_EXCLUDES="/home/*/*"
#   BACKUP_EXTRA_INCLUDES="/home/*/.config /home/*/.ssh /home/*/.gnupg /home/*/.local/share/keyrings"
BACKUP_EXTRA_INCLUDES=""

# Capacity policy: refuse (default) | warn | off. The floor is measured on the
# whole filesystem of each source, so on a btrfs system disk that also holds
# 1.5 TiB of VM images beside a 100 GiB system, "warn" lets a 1 TiB drive be
# used.
#CAPACITY_CHECK="refuse"

# How long a layer waits for another layer's run to finish (seconds; 0 = skip
# instead). The tray and the timers can start two layers at once.
#BX_LOCK_WAIT=7200