Julien LutranandClaude Fable 5 6ebc93db2e seafile-gc: diagnose never-run GC (24.9M inodes) and the reclamation procedure ...

GC never ran in the server's life: 13 ghost libraries, unlimited
history on 16/17 repos, 867G blocks vs 710G live data, and an
fs/commits object swarm that makes per-file backup walks take days.
Procedure: pre-GC zfs snapshot, keep_days=90 default, offline
seaf-gc (+ --rm-deleted, --rm-fs), measure, quarterly GC schedule,
then re-decide the S3 backup mechanism and the 13.x upgrade.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

2026-08-26 16:37:05 +02:00

..

incus-copy.md

Initial import: nuc and ks4 infrastructure documentation

2026-08-09 22:50:07 +02:00

install.md

Initial import: nuc and ks4 infrastructure documentation

2026-08-09 22:50:07 +02:00

local-backup-cron.md

Document the ks2 decommission plan and the new ks4 backup legs

2026-08-23 08:56:28 +02:00

plakar-incus-integration.md

plakar-incus-integration: analyze the giant-instance seek-bound wall

2026-08-24 22:23:09 +02:00

plakar-s3-data.md

plakar-s3-data: growth issue filed upstream as PlakarKorp/plakar#2338

2026-08-25 09:29:28 +02:00

restic-backup.md

restic-backup: generic exclude file name

2026-08-25 13:07:41 +02:00

seafile-gc.md

seafile-gc: diagnose never-run GC (24.9M inodes) and the reclamation procedure

2026-08-26 16:37:05 +02:00

seafile-gcid-fix.md

plakar-s3-data: seed done except seafile-fs (EINTR livelock, excluded)

2026-08-24 00:07:01 +02:00