azcomputerguru/claudetools
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

cascades: recover 4 docs dropped by the history-rewrite/repo-split

Browse Source 
The 2026-06-18 repo restructure (history rewrite + project->submodule split)
dropped these 4 Cascades files from the new clone. Copied byte-identical from
the pre-cutover claudetools.old clone (md5-verified):
- docs/network/network-optimization-master-plan.md
- docs/network/phase1-voice-qos-design.md
- reports/2026-06-18-voice-quality-diagnostic.md
- session-logs/2026-06/2026-06-18-howard-cascades-rf-voice-optimization-plan.md

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Howard Enos
2026-06-18 20:21:27 -07:00
parent b66b80a95b
commit c2e5f4faeb
4 changed files with 514 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

71
clients/cascades-tucson/reports/2026-06-18-voice-quality-diagnostic.md Normal file
View File

@@ -0,0 +1,71 @@
# Cascades — Voice Quality Diagnostic (post VLAN 30 cutover)
- **Date:** 2026-06-18 (Howard-Home / claude-main)
- **Trigger:** All phones migrated to isolated VOICE VLAN 30 to improve call quality; users report
**dropped calls, breaks in voice, reception issues.** This is the RF/quality assessment.
- **Data source:** live UniFi controller `/stat/sta` + USW-16-PoE `port_table`, 2026-06-18.
## Cutover status — COMPLETE
31 devices on VOICE (`10.0.30.0/24`): 8 AudioCodes (`.224-.231`), 22 Poly (`.202-.223`), Vertical
desktop (`.201`). AudioCodes required a full power-off/on (external-powered; not PoE -> UniFi
power-cycle is a no-op) before they re-DHCP'd.
## Headline finding
**The VLAN move gives separation + sets up QoS, but it does NOT by itself fix call quality.** The
dropped calls / voice breaks are an **RF problem on the WiFi (Poly) phones.** The wired AudioCodes
are clean. Quality fixes are RF + QoS, below.
## Wired AudioCodes (8) — HEALTHY
All USW-16-PoE ports 1-8: up, 100M full-duplex, **rx_err=0 tx_err=0 rx_drop=0.** No network-layer
problem. 100M is fine for voice. With VLAN isolation + QoS these desk phones should be solid.
## WiFi Poly phones — RF problems (retry% = the call-quality killer)
Thresholds: retry >10% = audible breaks; RSSI <-67 marginal, <-75 bad; voice wants 5 GHz.
**SEVERE — fix first:**
| Phone (IP) | User/Loc | AP | Band | RSSI | Retry | Issue |
|---|---|---|---|---|---|---|
| 10.0.30.202 | Lauren / Accounting | CC Bridge | **2.4** | -56 | **50%** | stuck on 2.4 GHz, half packets retransmit |
| 10.0.30.218 | Shelby / MemCare Dir | MemCare Nurse Stn | **2.4** | -56 | **53%** | stuck on 2.4 GHz |
| 10.0.30.220 | Christine / rm 515 | 517 | 5 | **-82** | 7% | coverage gap (signal near-unusable) |
| 10.0.30.219 | Karen Rossini / rm 515 | 517 | 5 | -75 | 16% | weak + high retry |
**MODERATE:**
| 10.0.30.212 | rm 204 | 204 | 5 | -74 | 13% | weak + retry |
| 10.0.30.213 | Medtech rm 206 | 206 U7 Pro | 5 | -66 | 13% | 5 GHz congestion |
| 10.0.30.214 / .215 | rm 210 | 210 | 5 | -72 | 7-9% | weak |
| 10.0.30.206 | Dining Room | Dining Room | 5 | -70 | 9% | borderline |
**Healthy (reference):** .207/.209/.221/.222/.223 etc. — 5 GHz, RSSI -41 to -60, retry <3%.
### Three root causes
1. **2.4 GHz with ~50% retry** (Lauren .202, Shelby .218) — the single worst issue; matches the
documented Cascades 2.4 GHz saturation. **Must force these to 5 GHz.**
2. **Coverage gaps** — rooms 515 (-82/-75), 210/204 (-72/-74): too far from the serving AP; weak
signal drops calls when RF varies or people move.
3. **5 GHz congestion** — several at 13-16% retry on 5 GHz (80 MHz width + channel overlap, per the
2026-06-16 audit).
## Stragglers — 6 Poly phones NOT on VOICE
Five on VLAN 20 (`10.0.20.64/.65/.66/.67/.195`) + one on `192.168.1.126`. `10.0.20.66` (Dining
Room) is at **35% retry.** Missed in cutover or still on the old PPSK key -> migrate to the voice
PPSK so all phones are isolated + benefit from voice QoS.
## Recommended fixes (prioritized; NONE applied — Cascades requires explicit per-change go)
1. **QoS for the voice VLAN (NEW capability the move enables) — highest ROI, lowest risk.** Mark
VLAN 30 voice traffic DSCP EF / priority on pfSense + UniFi so voice gets priority under load ->
reduces jitter/breaks network-wide.
2. **Force voice phones off 2.4 GHz** — on the CSCNet voice PPSK / the APs serving .202 & .218,
disable 2.4 GHz association for voice (or band-steer to 5/6 GHz). Fixes Lauren + Shelby (the two
worst) immediately.
- **DONE 2026-06-18: Lauren `.202` locked to AP 103** (off the CC Bridge wireless-mesh AP -> wired AP). **INTERDEPENDENCY:** AP 103's 5 GHz is saturated (ch149, 75% airtime, ~25,900 retries, 12 clients) -> tonight's 5 GHz plan MUST relieve AP 103 (channel off 149 / 80->40 MHz / load-balance) or she trades a mesh problem for a congestion problem.
- Shelby `.218` is floor 5/6 (MemCare) -> **out of scope tonight** per Howard.
3. **Coverage** — rooms 515, 210, 204: check AP placement/power; consider a closer AP or raising the
nearest AP's power; min-RSSI to push phones off far APs. (Ties into the staged coverage-thin /
2.4 remediation runbooks.)
4. **Migrate the 6 straggler phones** to the voice PPSK (VLAN 30).
5. **5 GHz width/channel** — apply the staged audit recommendation (40 MHz width, non-DFS plan) to
cut co-channel retry.
## Next
Discuss + pick changes. QoS (#1) + 2.4 GHz force-off (#2) are the fastest wins for the complaints.