feat(multiplayer): run leased allocator worker

This commit is contained in:
Josh Creek
2026-09-01 10:38:45 +01:00
parent 6f7d61eafb
commit 55c46f56ec
6 changed files with 255 additions and 6 deletions
+60
View File
@@ -0,0 +1,60 @@
package allocator
import (
"context"
"fmt"
"strconv"
"time"
"github.com/cosmic-clash/cosmic-clash/server/domain"
)
// MatchClaimSource is the durable allocator work queue. Implementations must
// lease a match before returning it and fence binding by allocation ID.
type MatchClaimSource interface {
ClaimAllocatingMatch(context.Context, time.Time) (domain.AllocationRequest, bool, error)
BindAllocatedMatch(context.Context, domain.Allocation) error
}
// Worker consumes one leased match at a time. Provider failures deliberately
// retain the lease: an HTTP/provider failure can be ambiguous after an external
// allocation, so releasing it could allocate two GameServers for one match.
type Worker struct {
Claims MatchClaimSource
Service Service
Now func() time.Time
}
// RunOnce returns whether it found a claimed match. It never exposes an
// endpoint itself; Service first records the provider allocation durably and
// BindAllocatedMatch then attaches that already-recorded allocation to the
// fenced match claim.
func (w Worker) RunOnce(ctx context.Context) (bool, error) {
if w.Claims == nil || w.Now == nil {
return false, errNotConfigured
}
request, found, err := w.Claims.ClaimAllocatingMatch(ctx, w.Now())
if err != nil || !found {
return found, err
}
result, err := w.Service.Allocate(ctx, request, AllocationLabels(request))
if err != nil {
return true, fmt.Errorf("allocate claimed match %s: %w", request.MatchID, err)
}
if err := w.Claims.BindAllocatedMatch(ctx, result.Allocation); err != nil {
return true, fmt.Errorf("bind allocated match %s: %w", request.MatchID, err)
}
return true, nil
}
// AllocationLabels are the compatibility selectors shared with the Fleet
// template. They are derived only from the durable match plan, never client
// input or mutable worker configuration.
func AllocationLabels(request domain.AllocationRequest) map[string]string {
return map[string]string{
"cosmic-clash.io/region": request.Region,
"cosmic-clash.io/build": request.Build,
"cosmic-clash.io/protocol": strconv.Itoa(request.Protocol),
"cosmic-clash.io/transport": request.Transport,
}
}
+69
View File
@@ -0,0 +1,69 @@
package allocator
import (
"context"
"errors"
"reflect"
"testing"
"time"
"github.com/cosmic-clash/cosmic-clash/server/agones"
"github.com/cosmic-clash/cosmic-clash/server/domain"
)
type matchClaimSpy struct {
request domain.AllocationRequest
found bool
err error
bound domain.Allocation
bindErr error
}
func (s *matchClaimSpy) ClaimAllocatingMatch(_ context.Context, _ time.Time) (domain.AllocationRequest, bool, error) {
return s.request, s.found, s.err
}
func (s *matchClaimSpy) BindAllocatedMatch(_ context.Context, allocation domain.Allocation) error {
s.bound = allocation
return s.bindErr
}
func TestWorkerClaimsAllocatesAndBindsDurably(t *testing.T) {
request := domain.AllocationRequest{AllocationID: "allocation-1", MatchID: "match-1", Region: "EU", Build: "build-1", Protocol: 1, Transport: "enet"}
claims := &matchClaimSpy{request: request, found: true}
provider := &providerSpy{result: agones.AllocatedServer{Allocation: domain.Allocation{AllocationID: request.AllocationID, MatchID: request.MatchID, ServerID: "server-1", Region: request.Region, Build: request.Build, Protocol: request.Protocol, Transport: request.Transport, State: domain.ServerAllocated}, Endpoint: "127.0.0.1:7777"}}
durable := &durableSpy{}
worker := Worker{Claims: claims, Service: Service{Provider: provider, Durable: durable, Now: func() time.Time { return time.Unix(1_000, 0) }}, Now: func() time.Time { return time.Unix(1_000, 0) }}
processed, err := worker.RunOnce(context.Background())
if err != nil || !processed || provider.calls != 1 || durable.calls != 1 || claims.bound.ServerID != "server-1" {
t.Fatalf("processed=%t err=%v provider=%d durable=%d bound=%+v", processed, err, provider.calls, durable.calls, claims.bound)
}
}
func TestWorkerRetainsClaimWhenProviderOutcomeIsAmbiguous(t *testing.T) {
request := domain.AllocationRequest{AllocationID: "allocation-1", MatchID: "match-1", Region: "EU", Build: "build-1", Protocol: 1, Transport: "enet"}
claims := &matchClaimSpy{request: request, found: true}
provider := &providerSpy{err: errors.New("provider timeout")}
worker := Worker{Claims: claims, Service: Service{Provider: provider, Durable: &durableSpy{}, Now: func() time.Time { return time.Unix(1_000, 0) }}, Now: func() time.Time { return time.Unix(1_000, 0) }}
processed, err := worker.RunOnce(context.Background())
if err == nil || !processed || claims.bound != (domain.Allocation{}) {
t.Fatalf("processed=%t err=%v bound=%+v", processed, err, claims.bound)
}
}
func TestWorkerDoesNothingWhenNoDurableMatchIsAvailable(t *testing.T) {
claims := &matchClaimSpy{}
worker := Worker{Claims: claims, Now: func() time.Time { return time.Unix(1_000, 0) }}
processed, err := worker.RunOnce(context.Background())
if err != nil || processed {
t.Fatalf("processed=%t err=%v", processed, err)
}
}
func TestAllocationLabelsMirrorFleetCompatibilityTuple(t *testing.T) {
got := AllocationLabels(domain.AllocationRequest{Region: "NA", Build: "build-4", Protocol: 12, Transport: "steam_sdr"})
want := map[string]string{"cosmic-clash.io/region": "NA", "cosmic-clash.io/build": "build-4", "cosmic-clash.io/protocol": "12", "cosmic-clash.io/transport": "steam_sdr"}
if !reflect.DeepEqual(got, want) {
t.Fatalf("labels=%v want=%v", got, want)
}
}
+83
View File
@@ -0,0 +1,83 @@
package main
import (
"context"
"database/sql"
"flag"
"log"
"os"
"os/signal"
"syscall"
"time"
"github.com/cosmic-clash/cosmic-clash/server/agones"
"github.com/cosmic-clash/cosmic-clash/server/allocator"
"github.com/cosmic-clash/cosmic-clash/server/migrations"
"github.com/cosmic-clash/cosmic-clash/server/store"
_ "github.com/jackc/pgx/v5/stdlib"
)
func main() {
dsn := flag.String("dsn", os.Getenv("COSMIC_CLASH_POSTGRES_DSN"), "PostgreSQL connection string")
migrationDir := flag.String("migrations", "migrations", "directory containing numbered SQL migrations")
agonesURL := flag.String("agones-url", os.Getenv("COSMIC_CLASH_AGONES_URL"), "Agones allocation API base URL")
namespace := flag.String("agones-namespace", envOrDefault("COSMIC_CLASH_AGONES_NAMESPACE", "default"), "Agones namespace")
transport := flag.String("transport", envOrDefault("COSMIC_CLASH_TRANSPORT", "enet"), "game transport: enet or steam_sdr")
interval := flag.Duration("interval", time.Second, "allocation poll interval")
flag.Parse()
if *dsn == "" || *agonesURL == "" {
fatalf("--dsn/COSMIC_CLASH_POSTGRES_DSN and --agones-url/COSMIC_CLASH_AGONES_URL are required")
}
if (*transport != "enet" && *transport != "steam_sdr") || *interval <= 0 {
fatalf("--transport must be enet or steam_sdr and --interval must be positive")
}
db, err := sql.Open("pgx", *dsn)
if err != nil {
fatalf("open PostgreSQL: %v", err)
}
defer db.Close()
startupCtx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
if err := db.PingContext(startupCtx); err != nil {
fatalf("ping PostgreSQL: %v", err)
}
if err := migrations.Apply(startupCtx, db, *migrationDir); err != nil {
fatalf("apply migrations: %v", err)
}
now := func() time.Time { return time.Now().UTC() }
worker := allocator.Worker{
Claims: store.AllocatingMatchClaims{DB: db, Transport: *transport},
Service: allocator.Service{
Provider: agones.Client{BaseURL: *agonesURL, Namespace: *namespace},
Durable: store.AllocationRegistry{DB: db},
Now: now,
},
Now: now,
}
ctx, stop := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
defer stop()
ticker := time.NewTicker(*interval)
defer ticker.Stop()
for {
if _, err := worker.RunOnce(ctx); err != nil && ctx.Err() == nil {
log.Printf("allocator: run once: %v", err)
}
select {
case <-ctx.Done():
return
case <-ticker.C:
}
}
}
func envOrDefault(name, fallback string) string {
if value := os.Getenv(name); value != "" {
return value
}
return fallback
}
func fatalf(format string, args ...any) {
log.Printf("allocator: "+format, args...)
os.Exit(1)
}
+34
View File
@@ -0,0 +1,34 @@
package store
import (
"context"
"database/sql"
"time"
"github.com/cosmic-clash/cosmic-clash/server/domain"
)
// AllocatingMatchClaims adapts the PostgreSQL lease boundary for allocator
// workers without making the allocator package depend on the store package.
type AllocatingMatchClaims struct {
DB *sql.DB
Transport string
}
func (s AllocatingMatchClaims) ClaimAllocatingMatch(ctx context.Context, now time.Time) (domain.AllocationRequest, bool, error) {
item, found, err := ClaimAllocatingMatch(ctx, s.DB, s.Transport, now)
return item.Request, found, err
}
func (s AllocatingMatchClaims) BindAllocatedMatch(ctx context.Context, allocation domain.Allocation) error {
return BindAllocatedMatch(ctx, s.DB, allocation)
}
// AllocationRegistry adapts provider-allocation reconciliation for allocator
// workers. A successful provider response is not publishable until this store
// boundary records the same compatibility tuple and GameServer identity.
type AllocationRegistry struct{ DB *sql.DB }
func (s AllocationRegistry) RecordProviderAllocation(ctx context.Context, allocation domain.Allocation, now time.Time) (domain.Allocation, error) {
return RecordProviderAllocation(ctx, s.DB, allocation, now)
}