Backup pattern: tangled-knot now, sturdy size-based rotation later #3

New issue

Closed

opened 2026-05-23 20:54:25 +00:00 by coilysiren · 1 comment

coilysiren commented

2026-05-23 20:54:25 +00:00

Owner

Originally filed by @coilysiren on 2026-05-23T01:42:23Z - https://github.com/coilysiren/infrastructure/issues/291

Tangled knot backup (immediate need)

The Tangled knot at tangled.coilysiren.me (deployed via #280, fronted via Traefik in #288) has no backup. Its data is /var/lib/tangled-knot - git repos plus the SQLite knotserver.db. Cheap to back up: a tar or snapshot of one directory.

For now, a knot-specific backup script or timer that tars /var/lib/tangled-knot. Naturally pairs with the autodeploy timer's backup-before-deploy sandwich (so deploys are safe) and a periodic backup independent of deploys.

Future: sturdy file-size-based backup pattern

The eventual goal is a really sturdy backup pattern across the homelab, but not today.

Direction: file-size-based rotation, not last-in-based. Instead of "keep last N backups," keep backups until the total backup directory size hits a threshold, then rotate the oldest out. Bounds disk use predictably across services with very different backup sizes.

When the general pattern lands, the tangled-knot backup and Forgejo's (see #269) both align to it. Until then, each service can ship a working-but-simple per-service backup script.

_Originally filed by @coilysiren on 2026-05-23T01:42:23Z - [https://github.com/coilysiren/infrastructure/issues/291](https://github.com/coilysiren/infrastructure/issues/291)_ ## Tangled knot backup (immediate need) The Tangled knot at tangled.coilysiren.me (deployed via #280, fronted via Traefik in #288) has no backup. Its data is `/var/lib/tangled-knot` - git repos plus the SQLite `knotserver.db`. Cheap to back up: a tar or snapshot of one directory. For now, a knot-specific backup script or timer that tars `/var/lib/tangled-knot`. Naturally pairs with the autodeploy timer's backup-before-deploy sandwich (so deploys are safe) and a periodic backup independent of deploys. ## Future: sturdy file-size-based backup pattern The eventual goal is a really sturdy backup pattern across the homelab, but not today. **Direction: file-size-based rotation, not last-in-based.** Instead of "keep last N backups," keep backups until the total backup directory size hits a threshold, then rotate the oldest out. Bounds disk use predictably across services with very different backup sizes. When the general pattern lands, the tangled-knot backup and Forgejo's (see #269) both align to it. Until then, each service can ship a working-but-simple per-service backup script.

coilysiren added the

label

2026-06-04 08:16:57 +00:00

coilysiren referenced this issue

2026-06-12 11:15:30 +00:00

forgejo runner pinned to offline kai-desktop-tower-wsl - docker-label CI queue starves #312

coilysiren referenced this issue

2026-06-12 11:21:46 +00:00

forgejo runner pinned to offline kai-desktop-tower-wsl - docker-label CI queue starves #312

coilysiren commented

2026-06-17 08:23:03 +00:00

Author

Owner

Backlog burndown 2026-06-17: closing low-priority (P3/P4) to bring the open count to a manageable level. Nothing lost — reopen if this resurfaces. Batch tag: burndown-2026-06.

Backlog burndown 2026-06-17: closing low-priority (P3/P4) to bring the open count to a manageable level. Nothing lost — reopen if this resurfaces. Batch tag: `burndown-2026-06`.

coilysiren

2026-06-17 08:23:03 +00:00