jcreek/CosmicClash
1
0
Fork 0
mirror of https://github.com/jcreek/CosmicClash.git synced 2026-09-10 16:04:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(multiplayer): report allocator readiness

Browse Source
This commit is contained in:
Josh Creek
2026-09-02 19:11:01 +01:00
parent 6ebd6e59c1
commit cbefa86c5c
6 changed files with 140 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
server/cmd/allocator/main.go
+15 -3
View File
@@ -26,6 +26,7 @@ func main() {
kubernetesTokenPath := flag.String("kubernetes-token-path", envOrDefault("COSMIC_CLASH_KUBERNETES_TOKEN_PATH", "/var/run/secrets/kubernetes.io/serviceaccount/token"), "rotating Kubernetes service-account bearer token")
kubernetesCAPath := flag.String("kubernetes-ca-path", envOrDefault("COSMIC_CLASH_KUBERNETES_CA_PATH", "/var/run/secrets/kubernetes.io/serviceaccount/ca.crt"), "Kubernetes API cluster CA bundle")
providerTimeout := flag.Duration("provider-timeout", 10*time.Second, "timeout for each Kubernetes/Agones API request")
readinessMaxStale := flag.Duration("readiness-max-stale", 30*time.Second, "maximum age of the last fully successful allocator cycle")
transport := flag.String("transport", envOrDefault("COSMIC_CLASH_TRANSPORT", "enet"), "game transport: enet or steam_sdr")
interval := flag.Duration("interval", time.Second, "allocation poll interval")
workloadSecret := flag.String("workload-secret", os.Getenv("COSMIC_CLASH_WORKLOAD_SECRET"), "HMAC secret for control-plane-issued workload tokens (see workload/signed_token.go); must match cmd/control-plane's own --workload-secret. Unset skips minting a cosmic-clash.io/workload-token annotation entirely")
@@ -36,8 +37,11 @@ func main() {
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 || *providerTimeout <= 0 {
fatalf("--transport must be enet or steam_sdr and --interval/--provider-timeout must be positive")
if (*transport != "enet" && *transport != "steam_sdr") || *interval <= 0 || *providerTimeout <= 0 || *readinessMaxStale <= 0 {
fatalf("--transport must be enet or steam_sdr and --interval/--provider-timeout/--readiness-max-stale must be positive")
}
if *readinessMaxStale < *interval+*providerTimeout {
fatalf("--readiness-max-stale must be at least --interval plus --provider-timeout")
}
if *allocationQuota < 0 || *allocationQuotaWindow <= 0 {
fatalf("--allocation-quota must be non-negative and --allocation-quota-window must be positive")
@@ -68,6 +72,7 @@ func main() {
log.Printf("allocator: enabled per-replica regional allocation quota=%d window=%s", *allocationQuota, *allocationQuotaWindow)
}
metrics := allocator.NewMetrics()
health := &allocator.Health{}
providerHTTP, err := agones.NewKubernetesHTTPClient(*agonesURL, *kubernetesTokenPath, *kubernetesCAPath, *providerTimeout)
if err != nil {
fatalf("configure Kubernetes API client: %v", err)
@@ -89,7 +94,7 @@ func main() {
defer stop()
var metricsServer *http.Server
if *metricsAddr != "" {
metricsServer = &http.Server{Addr: *metricsAddr, Handler: allocator.MetricsHandler(metrics)}
metricsServer = &http.Server{Addr: *metricsAddr, Handler: allocator.RoleHandler(metrics, health, *readinessMaxStale, now), ReadHeaderTimeout: 5 * time.Second}
go func() {
if err := metricsServer.ListenAndServe(); err != nil && err != http.ErrServerClosed {
log.Printf("allocator: metrics server: %v", err)
@@ -107,19 +112,26 @@ func main() {
ticker := time.NewTicker(*interval)
defer ticker.Stop()
for {
cycleHealthy := true
servers, err := client.ListReadyServers(ctx)
if err != nil && ctx.Err() == nil {
cycleHealthy = false
log.Printf("allocator: list Ready GameServers: %v", err)
} else {
for _, server := range servers {
if err := store.RegisterReadyServer(ctx, db, server, now()); err != nil && ctx.Err() == nil {
cycleHealthy = false
log.Printf("allocator: register Ready GameServer %s: %v", server.ServerID, err)
}
}
}
if _, err := worker.RunOnce(ctx); err != nil && ctx.Err() == nil {
cycleHealthy = false
log.Printf("allocator: run once: %v", err)
}
if cycleHealthy && ctx.Err() == nil {
health.ObserveSuccessfulCycle(now())
}
select {
case <-ctx.Done():
return