Migration complete:
* ana-docker on docker-ce 29.4.1, all 29 containers back up. Traefik
routing live (verified 200s on matrix.phasefinal.com presence +
seafile.phasefinal.com syncs).
* traefik-postboot.service installed + enabled on both traefik hosts
(esh-docker-vm, ana-docker) — one-shot systemd unit that restarts
traefik 60s after every boot, fixing the long-standing routing-races-
after-reboot symptom.
New playbook: remove-autorestic. Triggered by a typo (`D:escription`
in autorestic-backup.timer line 2) flagged by systemd-analyze during
the traefik-postboot install on esh-docker-vm. Rather than fix it,
remove autorestic — it's redundant with the PBS + structured-restic
two-layer pipeline that's been operational since 2026-04-22. Detected
on two ESH-side hosts: esh-docker-vm and esh-vm-db. Playbook removes
the four unit files + the /usr/local/bin/autorestic binary; leaves
/srv/backups/autorestic/.autorestic.yml (archival) and
/mnt/backup/restic/repo/esh (historical snapshots) for separate
disposition.
Sub-finding from ana-docker upgrade: seafile's seahub (the Python
frontend at port 8000 inside the container) failed to start because
mysql wasn't ready when seafile booted, and a single restart didn't
recover it. Traefik routes return 502 on seafile dynamic endpoints
until seahub is up. Needs separate triage of seafile's depends_on
wiring or seahub's retry behavior — not a docker-ce regression.
vh
ac282c5526