Files
llmgw/internal/control/adminapi/summary.go
T
Ilker ManapandClaude Opus 5 9d8cd5e284 Track which replicas are actually being used, and summarise the fleet
Two things a cluster with many nodes and many GPUs needs that were missing,
and they are the same thing seen from two distances.

Idle tracking. The control plane knew what was placed and whether it was
healthy, never whether anyone wanted it — so on a fleet with more models
than cards there was no way to answer the first question anyone asks, which
is what can I stop. The router is the only thing that knows, so it now
reports it: a timestamp written in Acquire, where a request already meets an
endpoint under a lock already held, and collected once a tick as a map. That
shape is the point. Per-request notification would be a firehose carrying
one bit — the newest timestamp wins — and this costs the same on four
replicas as on four hundred.

It travels through an interface in the direction opposite EndpointSink, for
the same reason that one exists: the control plane decides what should run,
the router knows whether anyone wants it, and neither should need to be in
the other's process to say so.

Unknown is kept distinct from idle throughout. A router that has just
restarted knows nothing, and a zero timestamp would be the oldest idle
replica in every comparison — so one restart would nominate an entire fleet
for eviction. Absent from the map, "-" in the table, negative in the JSON,
and IdleFor returns a second value rather than a duration anything could
act on by mistake.

ctl status. A per-node listing is right for four nodes and wrong for two
hundred: how full am I, what is broken, what is holding cards nobody is
using are aggregates, and reading them off a long table is a chore, which
means it gets skipped. Capacity per GPU class with usage bars, free whole
nodes per fabric — the number that decides whether a multi-node replica can
land, since free cards on the wrong fabric are not capacity for a model that
spans nodes — replica and model counts, models that cannot be placed, and
what has gone unasked-for. Derived from the same state the tables show, so
it cannot disagree with them.

Unhealthy GPUs count as neither used nor free, because counting them free is
how a cluster looks like it has room it does not.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-17 00:52:34 +02:00

190 lines
5.7 KiB
Go

package adminapi
import (
"sort"
"time"
)
// Summary is the whole fleet in one answer.
//
// A per-node listing is the right tool for four nodes and the wrong one for two
// hundred: the questions that matter at that size — how full am I, what is
// broken, what is holding cards it is not using — are aggregates, and reading
// them off a long table is a chore that gets skipped. Everything here is
// derived from the same state the tables show, so it can never disagree with
// them; it is a different question asked of the same facts.
type Summary struct {
Nodes NodeTotals `json:"nodes"`
GPUs []ClassUse `json:"gpus"`
// Fabrics matter because a multi-node replica may never span two of them,
// so free capacity in the wrong fabric is not capacity for the model that
// needs it.
Fabrics []FabricUse `json:"fabrics,omitempty"`
Replicas ReplicaTotal `json:"replicas"`
Models ModelTotals `json:"models"`
// Unplaceable names models the scheduler currently cannot satisfy. This is
// the line that turns "something is wrong" into a place to look.
Unplaceable []string `json:"unplaceable,omitempty"`
// IdleReplicas are serving replicas nothing has asked for in a while, which
// on a fleet with more models than GPUs is the first place to find room.
IdleReplicas []IdleReplica `json:"idle_replicas,omitempty"`
}
// NodeTotals counts nodes by the states that need different responses.
type NodeTotals struct {
Total int `json:"total"`
Ready int `json:"ready"`
Cordoned int `json:"cordoned"`
Draining int `json:"draining"`
Unreachable int `json:"unreachable"`
}
// ClassUse is how much of one kind of card is in use.
type ClassUse struct {
Class string `json:"class"`
Total int `json:"total"`
Used int `json:"used"`
// Unhealthy GPUs are neither used nor available, and counting them as free
// is how a cluster looks like it has room it does not have.
Unhealthy int `json:"unhealthy"`
}
// Free is what can still be allocated.
func (c ClassUse) Free() int { return c.Total - c.Used - c.Unhealthy }
// FabricUse is capacity within one fabric.
type FabricUse struct {
Fabric string `json:"fabric"`
Nodes int `json:"nodes"`
// LargestFreeBlock is the most nodes on this fabric that are entirely free,
// which is the number that decides whether a multi-node replica can land.
LargestFreeBlock int `json:"largest_free_block"`
}
// ReplicaTotal counts replicas by state.
type ReplicaTotal struct {
Ready int `json:"ready"`
Starting int `json:"starting"`
Draining int `json:"draining"`
Failed int `json:"failed"`
}
// ModelTotals counts models by whether they have what they asked for.
type ModelTotals struct {
Total int `json:"total"`
Satisfied int `json:"satisfied"`
Short int `json:"short"`
}
// IdleReplica is a replica holding hardware nothing is asking for.
type IdleReplica struct {
ID string `json:"id"`
Model string `json:"model"`
IdleSec int64 `json:"idle_seconds"`
GPUs int `json:"gpus"`
}
// idleThreshold is how long a replica must go unasked-for to be worth naming.
//
// Minutes rather than seconds, because a replica that flaps in and out of this
// list is noise, and because unloading one throws away its KV cache — the thing
// prefix routing spent its effort filling.
const idleThreshold = 10 * time.Minute
// Summary builds the fleet summary.
func (a *API) Summary() Summary {
var s Summary
nodes := a.Nodes()
byClass := map[string]*ClassUse{}
s.Nodes.Total = len(nodes)
for _, n := range nodes {
switch n.State {
case "ready":
s.Nodes.Ready++
case "cordoned":
s.Nodes.Cordoned++
case "draining":
s.Nodes.Draining++
default:
s.Nodes.Unreachable++
}
for _, c := range n.Classes {
if byClass[c] == nil {
byClass[c] = &ClassUse{Class: c}
}
}
// A node reports one class in practice; attribute its cards to it.
if len(n.Classes) == 1 {
u := byClass[n.Classes[0]]
u.Total += n.GPUTotal
u.Unhealthy += len(n.Unhealthy)
u.Used += n.GPUTotal - n.GPUFree - len(n.Unhealthy)
}
}
for _, u := range byClass {
s.GPUs = append(s.GPUs, *u)
}
sort.Slice(s.GPUs, func(i, j int) bool { return s.GPUs[i].Class < s.GPUs[j].Class })
byFabric := map[string]*FabricUse{}
for _, n := range nodes {
for _, f := range n.Fabrics {
if byFabric[f] == nil {
byFabric[f] = &FabricUse{Fabric: f}
}
byFabric[f].Nodes++
// Entirely free, which is what a multi-node replica needs: a node
// with one card in use cannot contribute a full-width shard.
if n.State == "ready" && n.GPUFree == n.GPUTotal && n.GPUTotal > 0 {
byFabric[f].LargestFreeBlock++
}
}
}
for _, f := range byFabric {
s.Fabrics = append(s.Fabrics, *f)
}
sort.Slice(s.Fabrics, func(i, j int) bool { return s.Fabrics[i].Fabric < s.Fabrics[j].Fabric })
for _, r := range a.Replicas("") {
switch r.State {
case "ready":
s.Replicas.Ready++
case "draining":
s.Replicas.Draining++
case "failed":
s.Replicas.Failed++
default:
s.Replicas.Starting++
}
if r.State == "ready" && r.IdleSec >= int64(idleThreshold.Seconds()) {
gpus := 0
for _, sh := range r.Shards {
gpus += len(sh.GPUs)
}
s.IdleReplicas = append(s.IdleReplicas, IdleReplica{
ID: r.ID, Model: r.Model, IdleSec: r.IdleSec, GPUs: gpus,
})
}
}
sort.Slice(s.IdleReplicas, func(i, j int) bool {
return s.IdleReplicas[i].IdleSec > s.IdleReplicas[j].IdleSec
})
for _, m := range a.Models() {
s.Models.Total++
if m.Ready >= m.Min {
s.Models.Satisfied++
continue
}
s.Models.Short++
// Short of replicas and unable to place more is the actionable case;
// short and still starting is just a cluster doing its job.
if m.Total <= m.Ready {
s.Unplaceable = append(s.Unplaceable, m.Name)
}
}
sort.Strings(s.Unplaceable)
return s
}