From f7c39c1700c3937cfb8799d1d0f6128be18bffb0 Mon Sep 17 00:00:00 2001 From: kisfenyo Date: Mon, 5 Oct 2026 22:03:52 +0200 Subject: [PATCH] =?UTF-8?q?CHANGELOG:=20unreleased=20=E2=80=94=20burn-down?= =?UTF-8?q?=20night=20ctrl-b=20lane=20(R-729/R-545,=20R-621,=20R-839,=20R-?= =?UTF-8?q?271,=20R-521,=20R-607/R-615,=20R-25)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Co-Authored-By: Claude Opus 5.5 (1M context) Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS --- CHANGELOG.md | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index b31c72f..440a1de 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,3 +1,15 @@ +## unreleased + +- R-729 + R-545: „Távoli mentési cél törlése" on the Távoli mentés page (POST /backup/offbox/clear, reveal-then-confirm) — the household can remove its own off-site target; the SSH key and known-host line are deleted, nothing on the target is touched; the repository password is kept whenever the hub holds a sealed package, the escrow is confirmed or the target ever held a backup. Refused for the Felhom-provided tier, during a running backup operation, and during an abandonment countdown. New strings (hu + en): `backups_remote.tavoli_mentesi_cel_torlese, backups_remote.cel_torlese_magyarazat, backups_remote.cel_torlese_megerosites, flash.offbox.clear_needs_confirmation, flash.offbox.clear_hub_tier, flash.offbox.clear_busy, flash.offbox.clear_abandon_pending, flash.offbox.target_cleared (hu+en)`. +- R-621: a held app's logs (logs page, /api/stacks//logs, remote diagnostics) show the log the hold kept before stopping it, when the app has no live output. The hold panel itself does not show it yet. +- R-839: installing an app refuses a data location that is a folder inside a drive (it must be the drive); such a value made the boot sweep hold the app after every restart. Diagnosed as a drill-harness value on 9202. New strings (hu + en): `api.deploy.hdd_path_inside_drive (hu+en)`. +- R-271: after a controller restart, the controller-to-agent channel coming back up is reported when a down alert was sent before the restart (a marker file in the data dir); before, the alarm the restart was meant to fix never closed. +- R-521: an app stopped because its drive was unplugged no longer sends its own app_start_failed mail; the drive's storage_disconnected event already names it. The dashboard still lists the app. +- R-607: a catalog sync that moves the catalog but no installed app's files now refreshes the update badge's input at once and no longer says „nincs változás". +- R-615: changing the catalog address (git.repo_url) now takes effect — the box re-clones from the new repository instead of fetching the old one for ever; a changed token only updates the stored address. +- R-25 (controller half): drive set-up mounts the new filesystem by the UUID the agent verified it made, not by re-reading the device path; an older agent still works through the old path, logged as a warning. +- Decision (`09` §3 decision 133, decided by CC unattended — operator may reverse): removing an own off-site target keeps the repository password whenever anything could depend on it. + ## v0.297.0 — small fixes from the burn-down: plain refusals that name a route, honest countdowns, a locked off-site store named, no shared metadata, gofmt gated (24 rows) (2026-10-05) **MinAgent: 0.131.0** (unchanged). New household strings (hu + en): `note.offsite.fail_locked`,