Post-reboot audit PASS + night window extended to 22:00-07:00 (#610)

Verified after the 22:53 multiqueue bounce: NIC queues live at 4/4 and
2/2 (the whole point of the reboot), PVE config clean with nothing
pending, both gateway lanes healthy, all 9 Open Terminal ports bound on
the tailscale IP, PSI metrics fresh, and no removed relaunchers
respawned. Moved the day-profile flip from 05:00 to 07:00 per human
ruling (live timer re-armed + staged copy and docs synced; the gateway
z.ai peak ladder is untouched). Also staged a ready-to-run script that
drops each account's own Open Terminal key into its ~/.creds as
openwebui.creds - not run yet; the master key list stays at
~/.creds/open-terminal.env.

💘 Generated with Crush

Assisted-by: Crush:glm-5.2

[#610]
This commit is contained in:
2026-09-01 07:30:59 -05:00
parent c810942239
commit b08d354a51
10 changed files with 131 additions and 20 deletions
+5 -5
View File
@@ -33,7 +33,7 @@ boot-race fix; the hardware half (~$300 parts) moved to ~Oct 2026, #601.]
| 2× compute GPU (models pending, Q5) | ~$300-400 if 2× RTX A2000 12GB | CUDA pool, no PSU change (75W each) |
Expected end state: dev VM at 20 vCPU / 128G with cgroup-fenced account lanes,
gateway and PMO dispatch permanently protected, nightly 22:00-05:00 batch burn
gateway and PMO dispatch permanently protected, nightly 22:00-07:00 batch burn
window aligned with the z.ai trickle ladder, sectestbed fleet subordinate by
host-side CPU weights, SDRs on their own controller + RT cores, and PSI-driven
backpressure wired end to end. All software pieces are staged and $0.
@@ -82,8 +82,8 @@ backpressure wired end to end. All software pieces are staged and $0.
8. Root growth to 500G is a safe online operation (MBR shuffle, §4.3).
[STATUS: pending — manual runbook, #603.]
9. Day/night dynamic profile (§5.5): staged timers flip batch weight, CPU pool,
memory fences and /data2 readahead at 22:00 and 05:00, matching your sleep
window (22:00-05:00) with the gateway's z.ai peak ladder (01:00-05:00 CST).
memory fences and /data2 readahead at 22:00 and 07:00, matching your sleep
window (22:00-07:00) with the gateway's z.ai peak ladder (01:00-05:00 CST).
Guest has no CPU governor to remove (KVM: host owns P-states; set host
governor to performance once, done).
[STATUS: LANDED — day/night timers live; interim values are the operating
@@ -428,11 +428,11 @@ beats the retired cop :8110 (#585) it still polls.
See §3.8; skeleton staged, token + PROTECTED list via Q12.
### 5.5 Day/night dynamic profile + PMO dispatch backpressure
Windows (Q14): OS night profile 22:00-05:00 local (your sleep window); the
Windows (Q14): OS night profile 22:00-07:00 local (your sleep window); the
z.ai peak ladder stays 01:00-05:00 CST as the LLM-trickle subset
(gateway/config.yaml:21-26, harness mirror config.go:235-244). Net effect:
22:00 the system opens up for batch, 01:00 LLM drops to trickle while CPU/IO
burn maxes, 05:00 everything returns to day bias.
burn maxes, 07:00 everything returns to day bias.
What flips at night (systemctl set-property --runtime + sysctl, all
non-persistent; a reboot lands safely in day mode; values in