Retention, Forget, And Multiple Repositories

[erenfro]

So, here's a problem I've now run into, and glad I caught it now, because it was going to become a serious issue due to overages and space issues.

I have a setup that backs up to my local in-house NAS, which then takes that repository and copies to BorgBase for onsite+offsite backups.

Problem is, regardless of how the setup is, my backups offsite in borgbase never follow any kind of retention policies at all, because it's only ever run during backup. And if I manually run forget on that profile, it never applies my specific retention policies, and expects me to supply the --keep-X Y parameters myself.

From this configuration sample, with sensitive bits sensored:

version = "1"

[global]
  # ionice is available on Linux only
  ionice = false
  ionice-class = 2
  ionice-level = 6
  priority = "low"
  default-command = "snapshots"
  initialize = false
  min-memory = 100

# a group is a profile that will call all profiles one by one
[groups]
  full-backup = [ "system" ]

# Default profile when not specified on the command line (-n or --name)
# There's no default inheritance from the 'default' profile,
# you can use the 'inherit' flag if needed
[default]
  repository = "rest:http://psql1:PASSWORD@HOST:8765/psql1/"
  #insecure-tls = true
  password-file = "key"
  initialize = false

  [default.env]
    TMPDIR = "/tmp"

  [default.retention]
    before-backup = false
    after-backup = true
    keep-hourly = 2
    keep-daily = 14
    keep-weekly = 26
    keep-monthly = 4
    keep-within = "3h"
    keep-tag = [ "forever" ]
    #compact = true
    prune = true
    tag = true
    host = true

# New profile named 'home'
[system]
  inherit = "default"
  initialize = false

  [system.backup]
    exclude-file = [ "excludes" ]
    exclude-if-present = [ ".nobackup" ]
    exclude-caches = true
    one-file-system = false
    tag = [ "system" ]
    source = [
        "/root",
        "/etc",
        "/var/log",
        "/usr/local",
        "/var/spool/cron",
        "/var/backups",
        "/srv",
        "/opt"
    ]
    no-error-on-warning = true
    schedule = "05:00"
    schedule-permission = "system"
    schedule-lock-wait = "2h"
    run-before = [
      "/etc/resticprofile/scripts/os before",
      "/etc/resticprofile/scripts/postgresql before",
    ]
    # Copy to Borgbase:
    run-after = [
      "/etc/resticprofile/scripts/postgresql after",
      "/etc/resticprofile/scripts/os after",
      "resticprofile system.copy"
    ]
    # On Failure:
    run-fail = [
      "/etc/resticprofile/scripts/postgresql fail",
      "/etc/resticprofile/scripts/os fail"
    ]
    # Clean up:
    run-finally = [
      "/etc/resticprofile/scripts/postgresql final",
      "/etc/resticprofile/scripts/os final"
    ]

    lock = "/tmp/resticprofile-system.backup.lock"

  [system.retention]
    before-backup = false
    after-backup = true
    keep-hourly = 2
    keep-daily = 14
    keep-weekly = 26
    keep-monthly = 12
    keep-within = "3h"
    keep-tag = [ "forever" ]
    prune = false
    tag = true
    host = true

  [system.prune]
    schedule = "Sun 10:00"
    schedule-permission = "system"
    schedule-lock-wait = "2h"

  [system.copy]
    after-backup = true
    initialize = true
    initialize-copy-chunker-params = true
    repository = "rest:https://SENSORED.borgbase.com"
    password-file = "key"
    lock = "/tmp/resticprofile-system.copy.lock"

[borgbase]
  inherit = "default"
  initialize = false
  repository = "rest:https://SENSORED.borgbase.com"

  [borgbase.retention]
    before-backup = false
    after-backup = true
    keep-hourly = 2
    keep-daily = 14
    keep-weekly = 26
    keep-monthly = 4
    keep-within = "3h"
    keep-tag = [ "forever" ]
    compact = true
    prune = true
    tag = true
    host = true

At present, when system.copy executes, nothing ever initiates a forget on retention, and if I run resticprofile borgbase.forget, it's expecting parameters to run as well, making this process much more tedious.

[erenfro]

So, I added under borgbase, a new section:

  [borgbase.forget]
    keep-hourly = 2
    keep-daily = 14
    keep-weekly = 26
    keep-monthly = 4
    keep-within = "3h"
    keep-tag = [ "forever" ]
    compact = true
    prune = true
    tag = true
    host = true

But I get a problem with this on --tag:

2024/12/16 14:27:27 using configuration file: profiles.toml
2024/12/16 14:27:27 profile 'borgbase': starting 'forget'
flag needs an argument: --tag
2024/12/16 14:27:27 forget on profile 'borgbase': exit status 1

Commenting out tag, does get around this issue, so perhaps a little bug in this area?