localai-org-maint-bot and Ettore Di Giacinto
c0b7e64973
feat(ui): show running models and host gauges on single-node installs ( #12189 )
...
On a single-node install nothing in Operate listed the models loaded on
this machine or let an admin stop one. The System page that did was
retired in #11548 , and its replacements (the Nodes workbench) only work
in distributed mode. The Nodes page also mis-detected single-node mode:
the cluster routes are not registered there, so /api/nodes answers 404,
but only 503 was treated as "distributed off", which sent every
single-node install to the empty worker-registration card. The rail hid
the entry anyway.
Nodes route on a single node becomes "This machine":
- the Nodes page's VRAM / RAM / CPU / models-disk gauges, fed from this
host by mapping /api/resources onto the worker heartbeat fields
- a memory bar splitting host RAM by running model
- a running-models table (backend, RSS, CPU share, uptime, PID) with
search, sorting, logs and a confirmed Stop
- the distributed setup behind an "Add machines" button
The Operate overview gains a "Running now" preview (heaviest five, with
Stop) on single node and a pointer to Nodes > Running models on a
cluster. The rail shows "This machine" in Runtime with a running count.
Backend, additive only:
- /system: each loaded model carries a `process` block (pid, rss_bytes,
memory_percent, cpu_percent, started_at). A sampler keeps one gopsutil
handle per PID so CPU is the share since the previous poll rather than
the lifetime average; it is omitted on the first reading.
- /api/resources: host `cpu` and models-path `disk`, the same readings
workers send in their heartbeat.
Also fixes the fleet tables widening the page on phones: the headers'
absolutely positioned sr-only labels escaped the scroll wrapper.
Assisted-by: Claude:claude-opus-5 [Playwright]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Co-authored-by: Ettore Di Giacinto <mudler@localai.io >
2026-09-21 19:12:35 +02:00
localai-org-maint-bot and Ettore Di Giacinto
997d403de4
feat(nodes): add fleet operations dashboard ( #12046 )
...
* feat(nodes): report CPU telemetry
Assisted-by: Codex:gpt-6
* feat(nodes): add fleet view utilities
Assisted-by: Codex:gpt-6
* feat(nodes): add fleet operations dashboard
Replace the panel roster with aggregate capacity gauges, fleet filtering and selection, bounded bulk actions, and an on-demand node inspector. Extend node details and distributed-mode documentation with CPU and models-disk telemetry.
Assisted-by: Codex:gpt-6
* fix(nodes): harden fleet lifecycle actions
Assisted-by: Codex:gpt-6
* fix(nodes): restore compact fleet composition
Keep fleet health, capacity, and attention in one compact overview at ordinary desktop widths. The inspector now overlays the roster until the workbench can preserve a useful table beside it.
Assisted-by: Codex:gpt-6
* feat(nodes): add accessible running models workbench
Assisted-by: Codex:gpt-6
* fix(nodes): correct model view ARIA links
Keep each tab panel available for its controlling tab while native hidden state removes inactive content from accessibility navigation. Model controls now expose only supported state and valid inspector relationships.
Assisted-by: Codex:gpt-6
* fix(nodes): align lifecycle and capacity states
Pending nodes now expose approval wherever node actions appear, while other lifecycle controls follow the server transition rules. Capacity totals exclude incomplete readings so missing availability remains unknown.
Assisted-by: Codex:gpt-6
* fix(nodes): restore approved dashboard composition
Assisted-by: Codex:gpt-6
* fix(nodes): integrate operate navigation
Assisted-by: Codex:gpt-6
* fix(nodes): restore low density fleet view
Assisted-by: Codex:gpt-6
* fix(nodes): preserve complete operate menu
Assisted-by: Codex:gpt-6
* fix(nodes): preserve inspector workspace height
Assisted-by: Codex:gpt-6
* fix(nodes): restore standard operate navigation
Assisted-by: Codex:gpt-6
* feat(ui): add collapsible console rail
Assisted-by: Codex:gpt-6
* feat(nodes): stop models from fleet view
Assisted-by: Codex:gpt-6
* fix(nodes): make inspector a full height drawer
Assisted-by: Codex:gpt-6
* fix(model): stop mixed local and remote placements
Assisted-by: Codex:gpt-6
* fix(ui): announce action menu navigation
Assisted-by: Codex:gpt-6
* fix(nodes): keep inspector within viewport
Assisted-by: Codex:gpt-6
* fix(ui): preserve focus across model actions
Assisted-by: Codex:gpt-6
---------
Co-authored-by: Ettore Di Giacinto <mudler@localai.io >
2026-09-15 00:57:14 +02:00
LocalAI [bot] and Ettore Di Giacinto
95b058e1c5
feat(ui): restructure Cluster Nodes view (pulse + panel roster + detail page) ( #10447 )
...
* chore: gitignore SDD scratch directory
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* feat(nodes): add GET /api/nodes/models cluster-wide loaded-models endpoint
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* feat(ui): add nodesApi.allModels() for cluster-wide model roster
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* feat(ui): move Scheduling to its own page and nav item
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* feat(ui): replace nodes stat-card strip with cluster pulse + attention callout
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* feat(ui): node-panel roster with inline model chips and segmented filter
Replace the Nodes table with a full-width node-panel roster that shows
each backend node's running-model chips without an expand click, plus an
All/Backend/Agent segmented filter. Per-node detail (models, backends,
labels, capacity) moves to the node detail page.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* feat(ui): add deep-linkable node detail page at /app/nodes/:id
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* fix(ui): remove em-dash from CapacityEditor comment; align detail spec backend mock
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* chore(ui): nodes page cleanup, hover/chip polish, docs for restructured cluster view
Nodes.jsx dead-code sweep confirmed clean (no StatCard/table/expand
state/scheduling-form leftovers). Two App.css polish fixes: move the
node-panel hover border-color onto the bordered element so hover gives
real feedback, and add the missing .model-chip__state rule the
ModelChip component already emits. Update distributed-mode docs prose to
describe the restructured cluster view (cluster pulse, attention
callout, node-panel roster with inline model chips, All/Backend/Agent
filter, node detail page at /app/nodes/:id, Scheduling as its own page).
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
* chore(ui): drop unused gpuVendorLabel export from nodeStatus
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Assisted-by: Claude:claude-opus-4-8 [Claude Code]
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io >
Co-authored-by: Ettore Di Giacinto <mudler@localai.io >
2026-06-22 18:24:29 +02:00
