SyslogSolution/prose-contracts
7
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

contracts: sync to ground truth — ornith-1.0-35b→strix-moe, 256K all GPUs, real LiteLLM timeouts

Browse Source 
Verified on ground 2026-07-16 against CT 116 litellm_config.yaml + GPU hosts:
- AMD host serves qwen3.6-35B-udq4 (LiteLLM alias strix-moe); ornith-1.0-35b does NOT exist
- All 3 GPUs at 256K ctx, parallel 2 (RTX 3090 was listed 128K/parallel 1)
- LiteLLM timeouts: qwen 300s, gemma 120s, strix 300s (were stale 90s/120s)
- Added LiteLLM model surface + key scoping to litellm-self-heal
- Patched health-check script path ref

Files: litellm-self-heal, litellm-health, gpu-fleet, gpu-self-heal,
zulip-adapter-lessons, abiba-zulip-restore, hermes-agent-baseline,
delegation-prose-contract, mumuni-delegation-prose-contract
This commit is contained in:
Abiba
2026-07-16 17:02:44 +00:00
parent d0feb7881e
commit dc572889f8
9 changed files with 49 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files
gpu-fleet.prose.md
+7 -7
View File
@@ -68,8 +68,8 @@ triggers:
│ RTX 3090 │ │ RTX 5070 │ │ Strix Halo│ │ GPU Monitor │
│ 24GB │ │ 12GB │ │ 64GB UMA │ │ :9100 │
│ 256K ctx │ │ 256K ctx │ │ 256K ctx │ │ Watchdog │
│ qwen3.6 │ │ gemma-4-12b │ │ ornith35B │ │ Prometheus │
│ 27B-code │ │ :8080 │ │ :8080 │ │ exporter │
│ qwen3.6 │ │ gemma-4-12b │ │ qwen3.6 │ │ Prometheus │
│ 27B-code │ │ :8080 │ │ -35B-udq4 │ │ exporter │
│ :8080 │ │ :9400 (exp) │ │ :9400(exp)│ │ :9401 │
│ :9400 │ └─────────────┘ └───────────┘ └──────────────┘
└──────────┘
@@ -103,9 +103,9 @@ but are deprecated for agent configs. Only the stable aliases survive model swap
| Model | GPU | Weight | RPM Cap | Timeout |
|-------|-----|--------|---------|---------|
| qwen3.6-27B-code | RTX 3090 (.8:8080) | **0.55** | 500 | 90s |
| qwen3.6-27B-code | RTX 3090 (.8:8080) | **0.55** | 500 | **300s** |
| qwen3.6-35B-udq4 | Strix Halo (.15:8080) | **0.30** | 60 | **300s** |
| gemma-4-12b | RTX 5070 (.110:8080) | **0.15** | 200 | **300s** |
| gemma-4-12b | RTX 5070 (.110:8080) | **0.15** | 200 | **120s** |
Note: All syslog-auto entries route directly to GPUs with `api_key: not-needed`. The router (port 9000) is NOT in the inference path.
@@ -186,7 +186,7 @@ Show full fleet status: GPUs, models, VRAM, context windows, parallel slots, act
3. Check LiteLLM: `curl http://192.168.68.116/health` (expect "I'm alive!")
4. Check LiteLLM models: `curl -H "Authorization: Bearer $MASTER_KEY" http://192.168.68.116/v1/models`
5. Check LiteLLM timeouts: `grep -n 'timeout:' /opt/inference-harness/litellm_config.yaml`
- gemma-4-12b: 120s, qwen3.6-27B: 90s, ornith-1.0-35b: 120s
- gemma-4-12b: 120s, qwen3.6-27B-code: 300s, qwen3.6-35B-udq4/strix-moe: 300s (ornith-1.0-35b does NOT exist — legacy name, do not use)
- global request_timeout: 300s, nginx proxy_read_timeout: 600s
6. Check AMD metrics: `curl http://192.168.68.15:9400/metrics` (Radeon 8060S, util%, VRAM, temp, power)
7. Check port conflicts: verify only one llama-server on :8080 per host
@@ -247,7 +247,7 @@ If no SSH access, send Zulip DM via abiba-bot with vault update instructions.
- **RTX 3090 runs `--parallel 2`** with MTP draft (spec-type draft-mtp, spec-draft-n-max 2).
- **RTX 3090 config**: `-c 262144 -ctk turbo4 -ctv turbo4 --parallel 2 --flash-attn on --cont-batching --spec-type draft-mtp`. Context corrected to 256K (2026-07-15). VRAM: 90%. Service: `/home/llmuser/llama-wrapper.sh`.
- **RTX 5070 config (2026-07-15)**: Switched to IQ4_NL + MTP draft (Q8_0) at 256K context. Gen speed: 122 tok/s (was 70). VRAM: 10.8/12.2GB (88%). No draft model pre-upgrade due to VRAM constraints. Service: `/home/llmuser/llama-wrapper.sh`. Config: `--model gemma-4-12b-it-IQ4_NL.gguf --spec-draft-model gemma-4-12b-it-Q8_0-MTP.gguf --spec-type draft-mtp --spec-draft-n-max 4 --ctx-size 262144`.
- **LiteLLM timeout tuning (2026-07-08)**: gemma-4-12b 25→120s, qwen3.6-27B-code 40→90s, syslog-auto (qwen route) 40→90s. Nginx proxy_read_timeout: 600s. Global request_timeout: 300s. Config at `/opt/inference-harness/litellm_config.yaml`.
- **LiteLLM timeout tuning (verified 2026-07-16 against `/opt/inference-harness/litellm_config.yaml` on CT 116)**: gemma-4-12b 120s, qwen3.6-27B-code 300s, qwen3.6-35B-udq4 300s, strix-moe 300s, syslog-auto routes all 300s. Nginx proxy_read_timeout: 600s. Global request_timeout: 300s.
- **Strix Halo GPU**: Vulkan is the working backend (ROCm/HIP path abandoned — HSA runtime blocked on Debian 13). Build at `/root/llama.cpp/build-vk/`, commit `4fc4ec5` (2026-07-01), ggml 0.15.3 shared-lib arch. Mesa RADV 25.0.7, KHR_coopmat fast path active. ~70 tok/s gen, 532 tok/s prompt. Service: `strix-server.service` on port 8080 (was `ornith-server.service`), model changed to `unsloth/Qwen3.6-35B-A3B-MTP-GGUF` (UD-Q4_K_M), alias `qwen3.6-35B-udq4`, 256K context, flash-attn + q8 KV. MTP support enabled for 1.4-2.2x faster inference.
- **Port conflict detection (2026-07-05)**: All 3 GPU wrappers now detect ghost processes squatting port 8080 before starting. `.8` and `.110` use inline pre-start check in `llama-wrapper.sh`; `.15` uses `/usr/local/bin/port-cleanup.sh` ExecStartPre. Replaces the blanket `pkill -9 -x llama-server` on .15 which would kill ALL llama-server instances regardless of port. Ghost detection was the root cause of .8 crash-looping for 27+ restarts (stale pid 25836 squatting 8080 after OOM kill).
- **Strix Halo thermal safeguard (2026-07-02)**: `strix-server.service` has `-n 8192` (hard generation cap per request). Without it, `--predict` defaults to -1 (infinity) — a runaway request from .123 (Mumuni) decoded 39,868 tokens over 24 min, pushing Tctl to 98°C (crit 89.8°C) and throttling 70→29 t/s. The cap bounds worst-case generation to ~5 min. Do NOT remove `-n` without a replacement ceiling. Sustained load hits ~84°C even at 92s; the APU is fanless/low-flow. Clients MUST also set `max_tokens`.
@@ -262,7 +262,7 @@ If no SSH access, send Zulip DM via abiba-bot with vault update instructions.
| GPU | Model | Gen tok/s | Prompt tok/s | Baseline | Context |
|-----|-------|-----------|--------------|----------|---------|
| RTX 3090 (.8) | qwen3.6-27B-code (MTP) | **63** | — | — | 131K |
| RTX 3090 (.8) | qwen3.6-27B-code (MTP) | **63** | — | — | **256K** |
| RTX 5070 (.110) | gemma-4-12b (IQ4_NL+MTP) | **191** | — | — | **256K** |
| Strix Halo (.15) | qwen3.6-35B-udq4 | **71** | — | — | **256K** |