Files
meshcore-analyzer/docs
n30nexandn30nex 8437d22356 docs: use consistent snapshots for backup and restore (#2345)
The deployment guide copied a running `meshcore.db`, which can omit
committed WAL data or produce an inconsistent backup. It now uses the
existing authenticated `/api/backup` snapshot, checks the download
before publishing it, and explains how to restore with all database
users stopped while preserving the old database and sidecars for
rollback.

Fixes #703, following the [maintainer's remaining documentation
scope](https://github.com/Kpa-clawbot/CoreScope/issues/703#issuecomment-6163256620).

- Protects API credentials in a header file and keeps incomplete
downloads out of the completed backup set.
- Links the separate account-backup guide; configuration and secrets
remain separate backup responsibilities.
- Replaces unsafe live-copy scheduling instructions with the same
checked download procedure.

Validation:
- Exact shell example with real curl/SQLite: valid download, HTTP
401/500, empty, HTML, corrupt database, and truncated transfer. Failures
preserved the previous backup and published no new `.db`.
- Existing `TestBackup*` passed. Chromium downloaded an authenticated
snapshot from a disposable instance; a row committed only in its
nonempty WAL survived and `quick_check` returned `ok`.
- Disposable stopped-file restore and rollback recovered the original
WAL-only row and preserved the original files byte-for-byte.
- `git diff --check` and privacy preflight passed.

TDD exemption: pure docs; no application behavior changed and no test
files modified. No new config/customizer settings. Live operator
services were not used for validation.

Co-authored-by: n30nex <6686206+n30nex@users.noreply.github.com>
2026-10-29 18:33:14 +01:00
..