alexion/dotfiles
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(nix): bound the store, generations and boot menu (task 0014)

Browse Source 
Garbage collection and store optimisation each run weekly, on separate
days so they never contend, and retained boot configurations are capped
at 15.
This commit is contained in:
alexion
2026-07-20 10:21:24 -04:00
parent dcc03155a2
commit c63c3079be
2 changed files with 37 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
.claude/tasks/0014-nix-store-housekeeping.md
View File

@@ -10,9 +10,21 @@ Retained boot configurations are capped at 15. Each generation stores a kernel a
## Acceptance criteria
- [ ] Automatic garbage collection is enabled weekly, deleting generations older than 30 days
- [ ] Store optimisation is scheduled weekly, rather than performed at build time
- [ ] Retained boot configurations are capped at 15
- [ ] These are declared as plumbing in the shared base config, so every future `Host` inherits them
- [ ] `nix flake check` builds the `neogaia` toplevel
- [x] Automatic garbage collection is enabled weekly, deleting generations older than 30 days
- [x] Store optimisation is scheduled weekly, rather than performed at build time
- [x] Retained boot configurations are capped at 15
- [x] These are declared as plumbing in the shared base config, so every future `Host` inherits them
- [x] `nix flake check` builds the `neogaia` toplevel
- [ ] Manual confirmation after a rebuild: the collection and optimisation timers exist and are scheduled
## Implementation Notes
The two schedules are named days rather than the bare `weekly` keyword.
systemd expands `weekly` to `Mon *-*-* 00:00:00`, which would have started collection and deduplication at the same instant every week, leaving `nix-optimise` hard-linking paths `nix-gc` was concurrently deleting.
Collection now runs `Mon 03:15` and optimisation `Thu 03:45`, which keeps both weekly and keeps them apart.
The boot configuration cap sits in the shared base as the task asks, and is inert rather than an error on a host that does not use systemd-boot.
A future host on another bootloader therefore inherits no cap, which is the one place the "every future host inherits them" promise does not reach.
The last criterion is left open: confirming the timers on a running system needs a `nixos-rebuild switch`, which needs a sudo credential this session did not have.
It closes the way tasks 0013, 0016 and 0017 did, in a follow-up commit once the operator has switched the machine and read `systemctl list-timers`.