PFVCluster/questions-v10.md

Questions — v10 (current round)

New round: O&M lane repo split + change management (work order 2026-09-03, #769). Still-open PMG items from v9 carried at the bottom. One-line answers fine.

Done while you were away (no answers needed — for the 20-min sitting)

• PFVCluster split into 12 dedicated KNEL repos (~/projects/KNEL/<repo>), each with Redmine ticket + Discourse topic + git history. Map: t/296 + AGENTS.md. PFVCluster keeps cluster content only; KNELIAC references the new repos.
• Unit tests moved with code (netinfra, facilities) and pass. All repos carry the enforcement layer; pre-push audits green.
• Change management: prod/preprod/sectestbed env map from live tsys5 state + snapshot lifecycle tool (scripts/pve-snapshot.sh, verified end-to-end on throwaway VM 59999, destroyed same session) — t/331. Audit log: t/298 reply #120. pfv-bms/netinfra/PMG/PBS hard-protected in the tool.

O&M round — new (one decision each)

OQ1 (repo visibility). Created all 12 repos PUBLIC per the 2026-08-31 ruling, matching PFVCluster's class. OK, or flip some/all private (one-click each; say which)?

OQ2 (legacy overlaps). Old repos TSYS-CA and KNELSecretsManager overlap the new ca / secrets. Archive them into the new scheme (redirect README), or leave as-is?

OQ3 (apt-satellite design). Recommend: start with a flat pinned-mirror (the #758 tsys-ca pattern), graduate to aptly when the fleet grows. Approve?

OQ4 (Prometheus placement). Recommend: Prometheus + Alertmanager on tsys-librenms alongside LibreNMS (same vantage, no new VM); Grafana stays on Cloudron. Approve, or name a host?

OQ5 (Kuma window automation). ~/.creds/uptime-kuma.env exists — may I wire it into pve-snapshot.sh so rollbacks auto-create the maintenance window (and close it after), instead of the current manual reminder?

OQ6 (preprod ownership). The 531xx preprod lane on tsys5: is THIS lane the driver for vendor-version soak (~1wk, then promote), with per-domain chats consulted? Or per-stack?

OQ7 (GLPI changes during converge loops). Until GLPI change-control is ruled (#705/#741): sectestbed playbook iterations stay Redmine-ticket-only, exception requests noted in the ticket. OK, or start logging GLPI Changes now?

OQ8 (pfv-bms → KNELBMS rename). Deferred per your golden rule (HA chat active). Confirm the rename lands with the HA chat once you bless it — this lane will not touch it.

Carried from v9 — still open (PMG, #696)

PQ5 (native PMG cluster?). Recommended: yes — config + quarantine replication between .1.11/.1.10, managed strictly serial. OK?

PQ6 (TLS from fleet CA #697?). Recommended: yes for the two admin UIs; SMTP stays opportunistic.

PQ7 (notify address). Confirm coo@turnsys.com for PMG alerts + held-mail notices?

PQ8 (VIP specifics). Confirm VIP 192.168.3.249 + smtp.knel.net + dhcpd host-decl pin (VRRP MAC 00:00:5e:00:01:0a)?

PQ9 (outbound cutover staging). Serial staged: PVE hosts + PBS first (verify a week), then VM postfix. OK?

PQ11 (Cloudron relay user — founder side). Create pfv-relay@turnsys.com for PMG submission, or point me at who does? Fallback: PMG delivers direct to recipient MXs.

PQ12 (k8s submission port). :25 unauthenticated from trusted LAN as default; :587+auth later if wanted. OK?

PQ10 (side findings — ticket placement). (a) tsys-ucs-02 root alias blackhole; (b) Wazuh alerting never configured (#335). Ride-alongs in #696 or separate sub-tickets?

Carried from v6 — still open

Q4. Second human approver in gitea? Blocks access-roster merges by design until landed. Does not block code/deploy work.