Skip to content

fix(1.10): correct a false claim and deepen cold-sec to 21 daily - #578

Merged
swares merged 1 commit into
mainfrom
fix/1-10-correct-the-dry-run-claim
Sep 6, 2026
Merged

fix(1.10): correct a false claim and deepen cold-sec to 21 daily#578
swares merged 1 commit into
mainfrom
fix/1-10-correct-the-dry-run-claim

Conversation

@swares

Copy link
Copy Markdown
Owner

The merged version of this change asserted that grouping by host,tags would remove nothing today. The dry-run showed otherwise: 17 removed at the primary and offsite, 22 at cold-sec. The error was reasoning about the policy in the abstract rather than against these snapshots -- under one merged series the union of daily-7/weekly-4/monthly-6 collapses to about twelve, and the primary held 29 only because each frozen path-set had private daily slots. Twenty-nine was the bug, not the baseline.

Two VHDX-bearing snapshots survive as the June and July monthlies, and restic dedupes, so either pins the full 224 GiB. No space is freed until they age out around Jan-Feb 2027, which vindicates this entry's original estimate over its 08-29 correction: the correction was right about the default grouping, the original about a sane one, and the entry disagreed with itself for a week over a flag neither version had checked.

copy_keep_daily 14 to 21: at 14, cold-sec would have dropped Aug 14, 15, 17, 18, 19 and 20, and the primary drops those too, so six days that exist today would have vanished from both local repos. Verified at 21 by dry-run -- keep 27, remove 16, and 16 is exactly the VMs-era count, so nothing from mid-August is lost. The primary was deliberately not raised instead, since deepening it past cold-sec would invert the hierarchy documented eleven lines above the value being changed. Offsite stays at 7/4/6 because R2 is the one repo where depth is billed.

The merged version of this change asserted that grouping by host,tags would remove nothing today. The dry-run showed otherwise: 17 removed at the primary and offsite, 22 at cold-sec. The error was reasoning about the policy in the abstract rather than against these snapshots -- under one merged series the union of daily-7/weekly-4/monthly-6 collapses to about twelve, and the primary held 29 only because each frozen path-set had private daily slots. Twenty-nine was the bug, not the baseline.
Two VHDX-bearing snapshots survive as the June and July monthlies, and restic dedupes, so either pins the full 224 GiB. No space is freed until they age out around Jan-Feb 2027, which vindicates this entry's original estimate over its 08-29 correction: the correction was right about the default grouping, the original about a sane one, and the entry disagreed with itself for a week over a flag neither version had checked.
copy_keep_daily 14 to 21: at 14, cold-sec would have dropped Aug 14, 15, 17, 18, 19 and 20, and the primary drops those too, so six days that exist today would have vanished from both local repos. Verified at 21 by dry-run -- keep 27, remove 16, and 16 is exactly the VMs-era count, so nothing from mid-August is lost. The primary was deliberately not raised instead, since deepening it past cold-sec would invert the hierarchy documented eleven lines above the value being changed. Offsite stays at 7/4/6 because R2 is the one repo where depth is billed.
@swares
swares merged commit c27459d into mainSep 6, 2026
4 checks passed
@swares
swares deleted the fix/1-10-correct-the-dry-run-claim branch September 6, 2026 20:54
swares added a commit that referenced this pull request Sep 6, 2026
…579)
#578 corrected BACKLOG.md and backup.yml but not this comment, which still asserted 'Nothing is deleted today'. It would have been written into the deployed unit file, where anyone next touching retention would read it.
forget --dry-run said otherwise before anything ran: primary keep 12 remove 17, cold-sec keep 27 remove 16, offsite keep 12 remove 17. The mistake was reasoning about the policy in the abstract rather than against these snapshots -- the primary held 29 only because each frozen path-set had private daily slots, so twenty-nine was the bug and not the baseline.
The part that was right is preserved: the June 29 and July 31 monthlies still carry the VHDX and restic dedupes, so either pins the full 224.7 GiB and no space is reclaimed until they age out around Jan-Feb 2027. 'Nothing is deleted' and 'no space is freed yet' are different statements and the comment conflated them.
Found because the phrase wraps across two lines, so a grep for it returned nothing -- the fourth pattern today that could not match what it was looking for.
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant

@swares