The 2026-09-26 same-day cutover that ended the Lucebox DS4.1 detour:
Dankserver's :7770 endpoint is now unsloth llama.cpp (the b11030-mix
prebuilt, ROCm gfx120X, bundled runtime) serving
unsloth/Qwen3.8-Flash-Next-GGUF at Q8_0 — 175.30 GiB across six sha256-
verified split parts — on both Radeon AI PRO R9700s, as a ROOT unit (the
frank-large shape: no container, no generated unit).
The deployed contract:
- Placement law: layer split -ts 1,1 with -ngl 999; the routed experts
and the 51B-param n-gram (PLE) table are COMPUTED in host RAM via
-ot exps=CPU,ngram_embeddings=CPU (the x4-pair measurement: streaming
over PCIe loses at 3.7 t/s, computing in RAM does not care).
per_layer_token_embd.weight=CPU is required by the shared MTP head's
borrow path (a GPU-resident borrow crashes the draft load).
mmap stays on; --cache-ram 16384.
- Spec chain draft-mtp,ngram-mod: the shared Q8_0 MTP head
(mtp-...-shared-Q8_0.gguf, 2,786,568,256 B) at depth 2, with the
draftless ngram-mod matcher chained behind it — the only legal way to
layer two spec implementations in one process.
- Vision explicit: --mmproj names the BF16 projector (this build never
autoloads; a server without it comes up healthy and text-only).
- Reasoning: --reasoning-format deepseek → reasoning_content; the
Qwen3.8 template's {xhigh, medium, low} reasoning_effort gate folded
by reasoningEffortMap in models.yml.
- One stream, the model's full 262,144 context; greedy server default
with Qwen-recommended sampling riding compat.extraBody.
The deployer's deploy_llm_primary is rewritten for the root unit: it
gates every artifact at its declared byte count (stat -c %s, never wc -c
— which read the whole 377 GiB lucebox file for minutes on 2026-09-26),
retires the lucebox user quadlet to dated home trash, waits out port
7770, relinks/enables/restarts the root unit and waits up to 1800 s for
/health. The quadlet-era tree under dankserver/frank-primary/ is
archived (dankserver/.AGENTTRASH/20260926-llamacpp-cutover/ on this
box; git history everywhere else) and its planner mapping replaced by
the scripts/ trio.
Observability flips back to the llamacpp namespace: a frank-primary
Prometheus job at 192.168.0.235:7770 (cross-host LAN — the netavark
published-port quirk that forces host.containers.internal for the two
Dankclaw endpoints is same-host only), the LlmEndpointDown runbook
text, and the dashboard description for three llama.cpp endpoints.
Docs: the new runbook (frank-primary-llamacpp-qwen38-flash-next-q8.md),
RETIRED banners on the lucebox and radiance runbooks and their index
entries, the AGENTS.md fleet section rewritten for the root-unit shape,
and the homepage frank-medium card renamed to frank-primary with the
Q8_0 Flash-Next spec (docs.html #models and the modality note updated:
images on frank-primary and frank-small, audio on frank-small only).
models.yml: the dankserver provider is "llamacpp (dankserver)" serving
frank-primary, contextWindow 262144, maxTokens 65536 (compaction-gap
law: 183,501 + 65,536 <= 262,144), input [text, image].
Suite: 346 green (frank_primary_unit.bats rewritten for the root unit —
16 tests pinning the placement law, the shared-head requirement, the
explicit projector, HIP env hygiene (BIND_HOST, never bare HOST), and
the deployer's byte-count gates; the quadlet-era planner and
changed-tests mappings rewritten for the trio; the observability scrape
test asserts three llamacpp jobs again).
Rollback: the lucebox image (localhost/lucebox-rocm:gfx1201) and the
378 GiB payload stay on the host; the Qwen payload stays fetched, so a
re-swap in either direction is a redeploy, not a refetch.
80 lines
3.3 KiB
Bash
80 lines
3.3 KiB
Bash
#!/usr/bin/env bats
|
|
# Seam: shared/homepage/services.json — the service config the checker reads
|
|
# and the pages render. It must be a JSON list; every entry carries name,
|
|
# category, section, and a host:port probe. category is user or admin;
|
|
# section must be one the page's SECTION_ORDER renders. method, when
|
|
# present, is tcp or http. wip, when set, is "open" or "closed".
|
|
|
|
load helpers
|
|
|
|
@test "services.json: valid schema" {
|
|
python3 - "$REPO_ROOT/shared/homepage/services.json" <<'PY'
|
|
import json, sys
|
|
path = sys.argv[1]
|
|
try:
|
|
data = json.load(open(path))
|
|
except FileNotFoundError:
|
|
sys.exit("services.json not found")
|
|
assert isinstance(data, list), "top level must be a list"
|
|
assert data, "the list must not be empty"
|
|
for svc in data:
|
|
for field in ("name", "category", "section", "probe"):
|
|
assert field in svc, f"{svc.get('name', '?')} is missing {field}"
|
|
assert svc["category"] in ("user", "admin"), svc["category"]
|
|
if "method" in svc:
|
|
assert svc["method"] in ("tcp", "http"), svc["method"]
|
|
wip = svc.get("wip")
|
|
assert wip is False or wip in ("open", "closed"), f"{svc['name']} wip must be false, 'open' or 'closed'"
|
|
host, _, port = svc["probe"].rpartition(":")
|
|
assert host and port.isdigit(), f"bad probe {svc['probe']}"
|
|
PY
|
|
}
|
|
|
|
@test "services.json: every section is one the page renders" {
|
|
# A service whose section is not in app.js's SECTION_ORDER for its
|
|
# category is silently dropped from the page. Pin the contract.
|
|
python3 - "$REPO_ROOT/shared/homepage/services.json" <<'PY'
|
|
import json, sys
|
|
data = json.load(open(sys.argv[1]))
|
|
allowed = {
|
|
"user": {"AI", "Web Apps", "Media Library Apps"},
|
|
"admin": {"Media Library Apps", "Observability", "AI", "Infrastructure"},
|
|
}
|
|
for svc in data:
|
|
assert svc.get("section") in allowed[svc["category"]], (
|
|
f"{svc['name']} has unrenderable section {svc.get('section')!r} for {svc['category']}")
|
|
PY
|
|
}
|
|
|
|
@test "services.json: every user service has an Open link, status-only backends do not" {
|
|
python3 - "$REPO_ROOT/shared/homepage/services.json" <<'PY'
|
|
import json, sys
|
|
data = json.load(open(sys.argv[1]))
|
|
for svc in data:
|
|
show = svc.get("showLink", True)
|
|
if show:
|
|
assert svc.get("url"), f"{svc['name']} shows a link but has no url"
|
|
else:
|
|
assert not svc.get("url"), f"{svc['name']} is status-only but has a url"
|
|
PY
|
|
}
|
|
|
|
@test "the LLM status cards track the fleet, not retired service names" {
|
|
# This drifted twice: the card probing Dankserver's :7770 was still named
|
|
# frank-small long after frank-medium took that endpoint, and the endpoint
|
|
# itself has since been frank-medium (radiance), frank-primary (lucebox)
|
|
# and frank-primary (llama.cpp). The probe target is what matters; the
|
|
# name has to follow it.
|
|
python3 - "$REPO_ROOT/shared/homepage/services.json" <<'PY'
|
|
import json, sys
|
|
services = json.load(open(sys.argv[1]))
|
|
by_probe = {s["probe"]: s["name"] for s in services}
|
|
primary = by_probe.get("192.168.0.235:7770")
|
|
assert primary == "frank-primary", f"Dankserver :7770 card is named {primary!r}"
|
|
# Dankclaw serves two endpoints on two ports: frank-large (unsloth llama.cpp
|
|
# on the iGPU, 7770) and frank-small (llama.cpp on the RTX 3090, 7771).
|
|
assert by_probe.get("127.0.0.1:7770") == "frank-large", by_probe.get("127.0.0.1:7770")
|
|
assert by_probe.get("127.0.0.1:7771") == "frank-small", by_probe.get("127.0.0.1:7771")
|
|
PY
|
|
}
|