CI / test (push) Failing after 1m31s
Disk was the one resource nothing reported. `statfs` was already being called on every model directory, but only to ask whether the filesystem was a network one — the free space it returns in the same struct was thrown away. So the only way to find out why a node was filling up was to ssh to it, which is what this system exists to avoid. Reported per filesystem rather than per configured directory. Free space belongs to the disk, not the path: two model directories on one disk are one pool of bytes, and counting them separately would let the control plane add 24 GB to 24 GB and place a 40 GB model on a node that has 24. Shared filesystems are summed separately from local ones and left out of a node's own total. Space on a filer belongs to everyone mounting it, so twenty nodes would otherwise each report the same free terabyte, and a scheduler would place twenty models into room for one. Nothing reported means nothing is known, everywhere: an agent too old to send storage and a node with no model directories both send nothing, and both must read as unknown rather than as a full disk. `ctl nodes` and the dashboard show a dash, the metric is absent rather than zero, and a missing directory — which is what an unmounted filer looks like — is skipped rather than reported as having no space. Shown as free out of total, because the first is useless without the second: 8 GB free is comfortable on a 10 TB filer and nearly full on a 10 GB root partition, and only the second stops a model being placed. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
408 lines
13 KiB
Go
408 lines
13 KiB
Go
// Package adminapi is the control plane's management surface, which `ctl`
|
|
// consumes (plan §11, phase 3).
|
|
//
|
|
// It is mounted on the admin listener rather than the client-facing one. The
|
|
// two have different audiences: one serves inference to whoever holds an API
|
|
// key, the other lets a caller cordon nodes and read the cluster's shape.
|
|
package adminapi
|
|
|
|
import (
|
|
"context"
|
|
"encoding/json"
|
|
"fmt"
|
|
"io"
|
|
"net/http"
|
|
"time"
|
|
|
|
"gitea.mynodes.xyz/ilker/llmgw/internal/control"
|
|
"gitea.mynodes.xyz/ilker/llmgw/internal/control/catalog"
|
|
"gitea.mynodes.xyz/ilker/llmgw/internal/control/inventory"
|
|
"gitea.mynodes.xyz/ilker/llmgw/internal/control/reconciler"
|
|
)
|
|
|
|
// API serves the management endpoints.
|
|
type API struct {
|
|
models Models
|
|
ctl *control.Controller
|
|
}
|
|
|
|
// New builds the API over a controller.
|
|
func New(ctl *control.Controller) *API { return &API{ctl: ctl} }
|
|
|
|
// Register mounts the endpoints on a mux under /api/v1.
|
|
func (a *API) Register(mux *http.ServeMux) {
|
|
mux.HandleFunc("GET /api/v1/summary", a.summary)
|
|
// Adding a model is a write, and the only one this API has. It lives here
|
|
// rather than on the client-facing listener for the same reason the rest of
|
|
// it does: reaching this port already means holding the control plane.
|
|
mux.HandleFunc("POST /api/v1/models", a.addModel)
|
|
mux.HandleFunc("DELETE /api/v1/models/{name}", a.removeModel)
|
|
mux.HandleFunc("GET /api/v1/nodes", a.listNodes)
|
|
mux.HandleFunc("GET /api/v1/models", a.listModels)
|
|
mux.HandleFunc("GET /api/v1/replicas", a.listReplicas)
|
|
mux.HandleFunc("GET /api/v1/explain-placement", a.explainPlacement)
|
|
mux.HandleFunc("POST /api/v1/nodes/{id}/state", a.setNodeState)
|
|
}
|
|
|
|
// --- responses ---
|
|
|
|
// NodeView is a node as `ctl nodes` shows it.
|
|
type NodeView struct {
|
|
ID string `json:"id"`
|
|
Hostname string `json:"hostname,omitempty"`
|
|
MgmtAddr string `json:"mgmt_addr,omitempty"`
|
|
State string `json:"state"`
|
|
Labels map[string]string `json:"labels,omitempty"`
|
|
|
|
GPUTotal int `json:"gpu_total"`
|
|
GPUFree int `json:"gpu_free"`
|
|
// MemoryGB is the machine's RAM, 0 when the agent did not report it. It is
|
|
// what decides whether a CPU profile can be placed here, so it belongs
|
|
// where someone looks when a model will not schedule.
|
|
MemoryGB int `json:"memory_gb,omitempty"`
|
|
// Models are the weights this node can see. Absent means it was not asked
|
|
// to look, which is not the same as holding none.
|
|
Models []string `json:"models_on_disk,omitempty"`
|
|
// DiskFreeGB and DiskTotalGB are the filesystems holding those weights,
|
|
// summed. Zero means the agent did not report any, which an older agent
|
|
// and a node with no model directories both do — so it reads as unknown
|
|
// rather than as a full disk.
|
|
//
|
|
// Summed because an operator asking "can this node take another model"
|
|
// wants one number; the per-filesystem detail is in the node's own record
|
|
// for anyone who needs it. Shared filesystems are left out of the sum: the
|
|
// space on a filer is not this node's to spend, and adding it would let a
|
|
// fleet of twenty nodes each believe it has the same free terabyte.
|
|
DiskFreeGB int `json:"disk_free_gb,omitempty"`
|
|
DiskTotalGB int `json:"disk_total_gb,omitempty"`
|
|
Classes []string `json:"gpu_classes,omitempty"`
|
|
Fabrics []string `json:"fabrics,omitempty"`
|
|
|
|
// Unhealthy lists GPUs kept out of placement and why, which is usually the
|
|
// answer when a node looks idle but nothing will schedule onto it.
|
|
Unhealthy []string `json:"unhealthy_gpus,omitempty"`
|
|
|
|
LastSeenAgo string `json:"last_seen_ago"`
|
|
AgentVersion string `json:"agent_version,omitempty"`
|
|
Replicas int `json:"replicas"`
|
|
}
|
|
|
|
// ModelView is a model as `ctl models` shows it.
|
|
type ModelView struct {
|
|
Name string `json:"name"`
|
|
Aliases []string `json:"aliases,omitempty"`
|
|
Source string `json:"source"`
|
|
Min int `json:"min_replicas"`
|
|
Max int `json:"max_replicas"`
|
|
Ready int `json:"ready_replicas"`
|
|
Total int `json:"total_replicas"`
|
|
Fallbacks []string `json:"fallbacks,omitempty"`
|
|
Profiles []string `json:"profiles"`
|
|
}
|
|
|
|
// ReplicaView is a replica as `ctl replicas` shows it.
|
|
type ReplicaView struct {
|
|
ID string `json:"id"`
|
|
Model string `json:"model"`
|
|
Profile string `json:"profile"`
|
|
State string `json:"state"`
|
|
Pool string `json:"pool,omitempty"`
|
|
Fabric string `json:"fabric,omitempty"`
|
|
Endpoint string `json:"endpoint,omitempty"`
|
|
AgeSec int64 `json:"age_seconds"`
|
|
// IdleSec is how long since a request was routed here. Negative means the
|
|
// router has not reported any use of this replica, which is not the same
|
|
// as idle — it is what a replica looks like before anything has asked for
|
|
// it, and after a router restart.
|
|
IdleSec int64 `json:"idle_seconds"`
|
|
Err string `json:"error,omitempty"`
|
|
Shards []ShardView `json:"shards"`
|
|
}
|
|
|
|
// ShardView is one shard of a replica.
|
|
type ShardView struct {
|
|
ID string `json:"id"`
|
|
Node string `json:"node"`
|
|
Rank int `json:"rank"`
|
|
GPUs []int `json:"gpus"`
|
|
State string `json:"state"`
|
|
Err string `json:"error,omitempty"`
|
|
}
|
|
|
|
// ExplainView is why a model could not be placed.
|
|
type ExplainView struct {
|
|
Model string `json:"model"`
|
|
// Placed is true when the model has replicas, in which case there is
|
|
// nothing to explain.
|
|
Placed bool `json:"placed"`
|
|
// Rejections is empty when placement was never attempted, which is itself
|
|
// the answer: the model is satisfied, or the reconciler has not run yet.
|
|
Rejections []RejectionView `json:"rejections,omitempty"`
|
|
}
|
|
|
|
// RejectionView is one rejected candidate.
|
|
type RejectionView struct {
|
|
Profile string `json:"profile"`
|
|
Pool string `json:"pool,omitempty"`
|
|
Node string `json:"node,omitempty"`
|
|
Reason string `json:"reason"`
|
|
}
|
|
|
|
// --- handlers ---
|
|
|
|
func (a *API) listNodes(w http.ResponseWriter, _ *http.Request) {
|
|
writeJSON(w, http.StatusOK, a.Nodes())
|
|
}
|
|
|
|
// Nodes builds the node view.
|
|
//
|
|
// Exported because the dashboard renders the same data server-side. Going back
|
|
// out through HTTP to fetch what is already in this process would add a way for
|
|
// the two to disagree.
|
|
func (a *API) Nodes() []NodeView {
|
|
replicasOn := map[string]int{}
|
|
for _, rep := range a.ctl.Reconciler().Replicas() {
|
|
for _, n := range rep.NodeIDs() {
|
|
replicasOn[n]++
|
|
}
|
|
}
|
|
|
|
inv := a.ctl.Inventory()
|
|
out := make([]NodeView, 0)
|
|
for _, n := range inv.Nodes() {
|
|
v := NodeView{
|
|
ID: n.ID, Hostname: n.Hostname, MgmtAddr: n.MgmtAddr,
|
|
State: n.State.String(), Labels: n.Labels,
|
|
GPUTotal: len(n.GPUs), GPUFree: len(inv.FreeGPUs(n.ID)),
|
|
MemoryGB: n.MemoryTotalMB / 1024,
|
|
Fabrics: n.FabricNames(), AgentVersion: n.AgentVersion,
|
|
Replicas: replicasOn[n.ID],
|
|
}
|
|
if free, total := n.LocalDisk(); total > 0 {
|
|
v.DiskFreeGB = int(free / (1 << 30))
|
|
v.DiskTotalGB = int(total / (1 << 30))
|
|
}
|
|
for _, w := range n.Weights {
|
|
v.Models = append(v.Models, w.Path)
|
|
}
|
|
seen := map[string]struct{}{}
|
|
for _, g := range n.GPUs {
|
|
if _, dup := seen[g.Class()]; !dup {
|
|
seen[g.Class()] = struct{}{}
|
|
v.Classes = append(v.Classes, g.Class())
|
|
}
|
|
if !g.Health.OK() {
|
|
v.Unhealthy = append(v.Unhealthy,
|
|
fmt.Sprintf("gpu %d: %s", g.Index, g.Health.Problems[0]))
|
|
}
|
|
}
|
|
if !n.LastSeen.IsZero() {
|
|
v.LastSeenAgo = time.Since(n.LastSeen).Round(time.Second).String()
|
|
}
|
|
out = append(out, v)
|
|
}
|
|
return out
|
|
}
|
|
|
|
func (a *API) listModels(w http.ResponseWriter, _ *http.Request) {
|
|
writeJSON(w, http.StatusOK, a.Models())
|
|
}
|
|
|
|
// Models builds the catalog view.
|
|
func (a *API) Models() []ModelView {
|
|
ready, total := map[string]int{}, map[string]int{}
|
|
for _, rep := range a.ctl.Reconciler().Replicas() {
|
|
total[rep.Model]++
|
|
if rep.State == reconciler.Ready {
|
|
ready[rep.Model]++
|
|
}
|
|
}
|
|
|
|
out := make([]ModelView, 0)
|
|
for _, m := range a.ctl.Catalog().Models() {
|
|
v := ModelView{
|
|
Name: m.Name, Aliases: m.Aliases, Source: m.Source,
|
|
Min: m.Scaling.Min, Max: m.Scaling.Max,
|
|
Ready: ready[m.Name], Total: total[m.Name],
|
|
Fallbacks: m.Fallbacks,
|
|
}
|
|
for _, p := range m.Profiles {
|
|
v.Profiles = append(v.Profiles, p.Name)
|
|
}
|
|
out = append(out, v)
|
|
}
|
|
return out
|
|
}
|
|
|
|
func (a *API) listReplicas(w http.ResponseWriter, r *http.Request) {
|
|
writeJSON(w, http.StatusOK, a.Replicas(r.URL.Query().Get("model")))
|
|
}
|
|
|
|
// Replicas builds the replica view, optionally filtered to one model.
|
|
func (a *API) Replicas(model string) []ReplicaView {
|
|
out := make([]ReplicaView, 0)
|
|
for _, rep := range a.ctl.Reconciler().Replicas() {
|
|
if model != "" && rep.Model != model {
|
|
continue
|
|
}
|
|
v := ReplicaView{
|
|
ID: rep.ID, Model: rep.Model, Profile: rep.Profile,
|
|
State: rep.State.String(), Pool: rep.Pool, Fabric: rep.Fabric,
|
|
Endpoint: rep.Endpoint, Err: rep.Err,
|
|
AgeSec: int64(time.Since(rep.CreatedAt).Seconds()),
|
|
IdleSec: -1,
|
|
}
|
|
if rep.LastUsedAt.IsZero() {
|
|
v.IdleSec = -1
|
|
} else {
|
|
v.IdleSec = int64(time.Since(rep.LastUsedAt).Seconds())
|
|
}
|
|
for _, s := range rep.Shards {
|
|
v.Shards = append(v.Shards, ShardView{
|
|
ID: s.ID, Node: s.NodeID, Rank: s.Rank,
|
|
GPUs: s.GPUIndices, State: s.State.String(), Err: s.Err,
|
|
})
|
|
}
|
|
out = append(out, v)
|
|
}
|
|
return out
|
|
}
|
|
|
|
// Models is the service that owns catalog changes, when there is one.
|
|
//
|
|
// Optional because a control plane with no store cannot persist a model, and
|
|
// pretending otherwise would accept a model that vanishes on restart.
|
|
type Models interface {
|
|
Add(ctx context.Context, m catalog.Model, by string) error
|
|
Remove(ctx context.Context, name string) error
|
|
}
|
|
|
|
// SetModels supplies the catalog-changing service.
|
|
func (a *API) SetModels(m Models) { a.models = m }
|
|
|
|
func (a *API) addModel(w http.ResponseWriter, r *http.Request) {
|
|
if a.models == nil {
|
|
writeError(w, http.StatusNotImplemented,
|
|
"this control plane has no store, so a model added here would not survive a restart")
|
|
return
|
|
}
|
|
var m catalog.Model
|
|
if err := json.NewDecoder(io.LimitReader(r.Body, 1<<20)).Decode(&m); err != nil {
|
|
writeError(w, http.StatusBadRequest, "the model definition is not valid JSON: "+err.Error())
|
|
return
|
|
}
|
|
if err := a.models.Add(r.Context(), m, r.Header.Get("X-Llmgw-Actor")); err != nil {
|
|
// Every failure here is the definition being wrong, which is the
|
|
// caller's to fix — a collision, a missing fallback, a profile that
|
|
// cannot be placed. None of them are server faults.
|
|
writeError(w, http.StatusBadRequest, err.Error())
|
|
return
|
|
}
|
|
writeJSON(w, http.StatusOK, map[string]string{"added": m.Name})
|
|
}
|
|
|
|
func (a *API) removeModel(w http.ResponseWriter, r *http.Request) {
|
|
if a.models == nil {
|
|
writeError(w, http.StatusNotImplemented, "this control plane has no store")
|
|
return
|
|
}
|
|
name := r.PathValue("name")
|
|
if err := a.models.Remove(r.Context(), name); err != nil {
|
|
writeError(w, http.StatusBadRequest, err.Error())
|
|
return
|
|
}
|
|
writeJSON(w, http.StatusOK, map[string]string{"removed": name})
|
|
}
|
|
|
|
func (a *API) summary(w http.ResponseWriter, _ *http.Request) {
|
|
writeJSON(w, http.StatusOK, a.Summary())
|
|
}
|
|
|
|
func (a *API) explainPlacement(w http.ResponseWriter, r *http.Request) {
|
|
model := r.URL.Query().Get("model")
|
|
if model == "" {
|
|
writeError(w, http.StatusBadRequest, "model is required")
|
|
return
|
|
}
|
|
if _, ok := a.ctl.Catalog().Resolve(model); !ok {
|
|
writeError(w, http.StatusNotFound, fmt.Sprintf("model %q is not in the catalog", model))
|
|
return
|
|
}
|
|
|
|
writeJSON(w, http.StatusOK, a.Explain(model))
|
|
}
|
|
|
|
// Explain builds the placement explanation for one model.
|
|
//
|
|
// It does not check that the model exists; the handler above does that so it
|
|
// can answer 404, while the dashboard only ever asks about models it just read
|
|
// out of the catalog.
|
|
func (a *API) Explain(model string) ExplainView {
|
|
v := ExplainView{Model: model}
|
|
for _, rep := range a.ctl.Reconciler().Replicas() {
|
|
if rep.Model == model && rep.State.Counts() {
|
|
v.Placed = true
|
|
break
|
|
}
|
|
}
|
|
if res, ok := a.ctl.Reconciler().Explain(model); ok {
|
|
for _, rej := range res.Rejections {
|
|
v.Rejections = append(v.Rejections, RejectionView{
|
|
Profile: rej.Profile, Pool: rej.Pool, Node: rej.Node, Reason: rej.Reason,
|
|
})
|
|
}
|
|
}
|
|
return v
|
|
}
|
|
|
|
// setNodeState implements cordon, uncordon and drain.
|
|
//
|
|
// One endpoint rather than three verbs, because they are the same operation
|
|
// with different targets and the difference matters: cordon stops new
|
|
// placement, drain also moves existing work off.
|
|
func (a *API) setNodeState(w http.ResponseWriter, r *http.Request) {
|
|
id := r.PathValue("id")
|
|
|
|
var body struct {
|
|
State string `json:"state"`
|
|
}
|
|
if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
|
|
writeError(w, http.StatusBadRequest, err.Error())
|
|
return
|
|
}
|
|
|
|
var st inventory.NodeState
|
|
switch body.State {
|
|
case "ready", "uncordon":
|
|
st = inventory.Ready
|
|
case "cordoned", "cordon":
|
|
st = inventory.Cordoned
|
|
case "draining", "drain":
|
|
st = inventory.Draining
|
|
default:
|
|
writeError(w, http.StatusBadRequest,
|
|
fmt.Sprintf("unknown state %q (want ready, cordoned or draining)", body.State))
|
|
return
|
|
}
|
|
|
|
if !a.ctl.Inventory().SetState(id, st) {
|
|
writeError(w, http.StatusNotFound, fmt.Sprintf("no node %q", id))
|
|
return
|
|
}
|
|
// Act on it now rather than at the next tick: an operator draining a node
|
|
// is usually about to do something to it.
|
|
a.ctl.Reconciler().Notify()
|
|
|
|
writeJSON(w, http.StatusOK, map[string]string{"id": id, "state": st.String()})
|
|
}
|
|
|
|
func writeJSON(w http.ResponseWriter, code int, v any) {
|
|
w.Header().Set("Content-Type", "application/json")
|
|
w.WriteHeader(code)
|
|
_ = json.NewEncoder(w).Encode(v)
|
|
}
|
|
|
|
func writeError(w http.ResponseWriter, code int, msg string) {
|
|
writeJSON(w, code, map[string]string{"error": msg})
|
|
}
|