The card described the machine and said nothing about the thing whose dashboard
it is. Cache size matters most: VV_CACHE_ROOT is under /tmp, which on Unraid is
RAM, so it is shown against the rootfs percentage it counts against. Jobs are
locks with a live process — four lock files were sitting there and one job was
actually running.
Twenty-four fetch chains ended in an empty catch, which is not error handling
but error deletion: the request fails, nothing renders, nothing is logged, and
the surface sits on Loading forever. That is how the mesh chat's ReferenceError
read as a slow load for as long as it did.
Local stays the default and the cheap path; mesh asks each partner for its own
sessions over SSH, live rather than cached, because a stream is true for
minutes and a cached one would be confidently wrong about the only thing the
card exists to say.
HOST2 has UHD Graphics 730 and reported no GPU at all. An iGPU has no VRAM,
no sensor of its own and no encode/decode split, so it is drawn from the
per-engine busy figures Intel actually publishes and the fields it lacks are
null rather than a confident zero.
mdNumDisabled counts a slot that was never populated, so HOST2 reported
"Emulating 1 disk" for a parity2 slot with no disk in it. Emulation is a
data-disk state, and an assigned identity is what separates a slot that lost
its disk from one that never had one.
A mirror was probing an Ollama it does not have and an index it never builds,
so the card read "unreachable / not built" about the wrong machine while the
model it actually uses was up. The card now names whose figures it is showing.
Curated state copied to every node is state that can disagree, so the index,
the model and the shared memory stay on the owner and each node reaches them
over the SSH trust onboarding already builds. Chats stay on the node that had
them; memory and bug reports stay the owner's to write.
Mesh is a mode rather than an AI profile — it has no model, tokens, reasoning or sources, so making it a profile would branch every profile-aware path on the one that has no model.
The old card totalled rsync's own logs, so it reported 'no data moved' across a link that had carried hundreds of gigabytes over SSH, the arr APIs, conf pushes and the Unraid API.
Profile order comes from the registry so sections keep their places as threads move, and the recents are capped to the same N rather than repeating everything above them.
A flat list buries 'where was I in General Chat' under whatever the tabs have been doing, since the busiest profile takes the top regardless of which card you are looking at.
Comments asserting a partner is down, being rebuilt or not yet installed are true on the day they are written and wrong afterwards, and nothing prompts anyone to revisit them. Behaviour is described generically; current state belongs in operator memory, not in tracked files.
It existed to render two fields that live in master.conf's HOST IDENTITIES group, which the shared renderer now includes — one settings surface for the page rather than two both editing the same file.
Windows with runs but no bytes say 'no data moved' rather than 0.00 GB, and rows before the byte-parser fix are excluded rather than summed into a false total.
The wizard had a Skip beside Create API Key and a second link that renamed itself Go to Scheduler/Partnership, so leaving early and finishing looked like the same action.
It was a browser flag set on click, and refreshing is the first thing anyone does when a page looks stuck — which put them back on the phase-1 panel the state exists to replace.
The mirror's own job ends in seconds while Phase 2 runs for minutes on the owner, so the screen sat unchanged on the phase-1 panel — which reads as a hang, and the reasonable response to a hang is to start clicking.
Step 1b ran the owner's SCRIPTS_ROOT path over SSH on the mirror, so it failed on every appdata install; the wizard poll expired at 5 minutes against a Phase 2 that takes over 4.
With nothing blocking but optional keys unfilled the banner rendered an empty string, so a finished onboard showed a blank panel beside a Join button still greyed at Running.
The phase-2 branch of the mirror panel returned the same Join partnership / Onboard block as the not-started state, so a finished onboard read as one that never ran.
The guard tested Tier 1 only, so onboard Step 1d's posture — Tier 1 open, every Tier 2 closed — made a healthy partner climb toward auto-offboard every 30 minutes. The counter holds intervals, not days, and was rendered raw.
Onboard now ends at Phase 2: connected, running, and saying so. Phase 3 sets the seeding gate posture and dispatches the seed. With MEDIA_SEED_ENABLED false there is no Phase 3 and the model is two phases.
A first seed is ~28 TB behind a 12.5 MB/s bwlimit, and it ran inline as Step 9d, so the
phase-2 flag every status reader depends on was written only after it finished.
Gmer4Lfe