Ilker ManapandClaude Opus 5.5
0d3eb82b9b
Nodes say where they are with llmgw.io/site, room, row, rack and pdu labels, from agent.yaml or `llmgw ctl label` (kept in the store, over the agent's). Placement spreads a model's replicas across racks and PDUs, keeps a multi-node replica within one domain when its profile says keep_within, places nothing in a domain held with `llmgw ctl facility hold` (optionally draining it), and refuses a shard that would take a domain past its power cap from facility.budgets, counting declared, derived and measured draw. Explain-placement names the domain, and keeps the rejections of a placed replica; `ctl nodes` gains LOCATION and --domain; `ctl facility` lists the domains; the console's Nodes view shows and sets location and holds. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
68 lines
2.1 KiB
Go
68 lines
2.1 KiB
Go
package facility
|
|
|
|
import (
|
|
"slices"
|
|
"testing"
|
|
)
|
|
|
|
func TestALocationIsReadFromTheWellKnownLabels(t *testing.T) {
|
|
loc := LocationOf(map[string]string{
|
|
"llmgw.io/site": "sto1", "llmgw.io/room": "hall-b", "llmgw.io/row": "12",
|
|
"llmgw.io/rack": "r12-07", "llmgw.io/pdu": "r12-07-a, r12-07-b,r12-07-a", "role": "gpu",
|
|
})
|
|
want := []Domain{"site:sto1", "room:hall-b", "row:12", "rack:r12-07", "pdu:r12-07-a", "pdu:r12-07-b"}
|
|
if got := loc.Domains(); !slices.Equal(got, want) {
|
|
t.Errorf("domains %v, want %v", got, want)
|
|
}
|
|
if loc.Short() != "r12-07 pdu r12-07-a,r12-07-b" {
|
|
t.Errorf("short %q", loc.Short())
|
|
}
|
|
if loc.At(Rack) != "rack:r12-07" || loc.At(PDU) != "pdu:r12-07-a" {
|
|
t.Errorf("at rack %q, at pdu %q", loc.At(Rack), loc.At(PDU))
|
|
}
|
|
}
|
|
|
|
func TestANodeWithoutLabelsIsNowhere(t *testing.T) {
|
|
loc := LocationOf(map[string]string{"role": "gpu"})
|
|
if !loc.Empty() || len(loc.Domains()) != 0 || loc.Short() != "" {
|
|
t.Errorf("an unlabelled node is somewhere: %+v", loc)
|
|
}
|
|
}
|
|
|
|
func TestADomainIsNamedAsLevelColonName(t *testing.T) {
|
|
d, err := ParseDomain(" Rack:r12-07 ")
|
|
if err != nil || d != "rack:r12-07" || d.Human() != "rack r12-07" {
|
|
t.Errorf("got %q, %v", d, err)
|
|
}
|
|
for _, bad := range []string{"r12-07", "rack:", "shelf:3"} {
|
|
if _, err := ParseDomain(bad); err == nil {
|
|
t.Errorf("%q was accepted as a domain", bad)
|
|
}
|
|
}
|
|
}
|
|
|
|
func TestTheLimitIsTheCapLessTheMargin(t *testing.T) {
|
|
p := Policy{Margin: 0.1, Budgets: map[Domain]Budget{"rack:r1": {CapWatts: 16000}}}
|
|
if got := p.Limit("rack:r1"); got != 14400 {
|
|
t.Errorf("limit %v, want 14400", got)
|
|
}
|
|
if p.Limit("rack:r2") != 0 {
|
|
t.Error("a domain with no budget has a limit")
|
|
}
|
|
}
|
|
|
|
func TestTheFirstHoldAmongANodesDomainsIsFound(t *testing.T) {
|
|
p := Policy{Holds: map[Domain]Hold{"pdu:p2": {Domain: "pdu:p2", Reason: "breaker"}}}
|
|
if h, ok := p.HeldIn([]Domain{"rack:r1", "pdu:p1", "pdu:p2"}); !ok || h.Reason != "breaker" {
|
|
t.Errorf("got %+v, %v", h, ok)
|
|
}
|
|
if _, ok := p.HeldIn([]Domain{"rack:r1"}); ok {
|
|
t.Error("an unheld rack is held")
|
|
}
|
|
c := p.Clone()
|
|
delete(c.Holds, "pdu:p2")
|
|
if _, ok := p.Holds["pdu:p2"]; !ok {
|
|
t.Error("changing a clone changed the original")
|
|
}
|
|
}
|