cozystack/cmd/cozystack-controller/main.go

/*
Copyright 2025.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/

package main

import (
	"crypto/tls"
	"errors"
	"flag"
	"fmt"
	"os"
	"time"

	// Import all Kubernetes client auth plugins (e.g. Azure, GCP, OIDC, etc.)
	// to ensure that exec-entrypoint and run can make use of them.
	_ "k8s.io/client-go/plugin/pkg/client/auth"

	corev1 "k8s.io/api/core/v1"
	"k8s.io/apimachinery/pkg/apis/meta/v1/unstructured"
	"k8s.io/apimachinery/pkg/fields"
	"k8s.io/apimachinery/pkg/runtime"
	"k8s.io/apimachinery/pkg/runtime/schema"
	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
	clientgoscheme "k8s.io/client-go/kubernetes/scheme"
	ctrl "sigs.k8s.io/controller-runtime"
	"sigs.k8s.io/controller-runtime/pkg/cache"
	"sigs.k8s.io/controller-runtime/pkg/client"
	"sigs.k8s.io/controller-runtime/pkg/cluster"
	"sigs.k8s.io/controller-runtime/pkg/healthz"
	"sigs.k8s.io/controller-runtime/pkg/log/zap"
	"sigs.k8s.io/controller-runtime/pkg/metrics/filters"
	metricsserver "sigs.k8s.io/controller-runtime/pkg/metrics/server"
	"sigs.k8s.io/controller-runtime/pkg/webhook"

	gatewayv1alpha1 "github.com/cozystack/cozystack/api/gateway/v1alpha1"
	gpuv1alpha1 "github.com/cozystack/cozystack/api/gpu/v1alpha1"
	internalv1alpha1 "github.com/cozystack/cozystack/api/internalapi/v1alpha1"
	cozystackiov1alpha1 "github.com/cozystack/cozystack/api/v1alpha1"
	"github.com/cozystack/cozystack/internal/controller"
	"github.com/cozystack/cozystack/internal/controller/cacert"
	"github.com/cozystack/cozystack/internal/controller/tenantgateway"
	"github.com/cozystack/cozystack/internal/controller/tenantlogrouting"
	"github.com/cozystack/cozystack/internal/controller/tenantquota"
	"github.com/cozystack/cozystack/internal/controller/wildcardsecret"
	"github.com/cozystack/cozystack/internal/gpu"
	gpucontroller "github.com/cozystack/cozystack/internal/gpu/controller"
	"github.com/cozystack/cozystack/internal/gpu/model"
	gpuwebhook "github.com/cozystack/cozystack/internal/gpu/webhook"
	"github.com/cozystack/cozystack/internal/telemetry"

	cmv1 "github.com/cert-manager/cert-manager/pkg/apis/certmanager/v1"
	helmv2 "github.com/fluxcd/helm-controller/api/v2"
	cosiv1alpha1 "sigs.k8s.io/container-object-storage-interface-api/apis/objectstorage/v1alpha1"
	gatewayv1 "sigs.k8s.io/gateway-api/apis/v1"
	gatewayv1alpha2 "sigs.k8s.io/gateway-api/apis/v1alpha2"
	gatewayv1beta1 "sigs.k8s.io/gateway-api/apis/v1beta1"
	// +kubebuilder:scaffold:imports
)

var (
	scheme   = runtime.NewScheme()
	setupLog = ctrl.Log.WithName("setup")
)

func init() {
	utilruntime.Must(clientgoscheme.AddToScheme(scheme))

	utilruntime.Must(cozystackiov1alpha1.AddToScheme(scheme))
	utilruntime.Must(gatewayv1alpha1.AddToScheme(scheme))
	utilruntime.Must(gpuv1alpha1.AddToScheme(scheme))
	utilruntime.Must(internalv1alpha1.AddToScheme(scheme))
	utilruntime.Must(gatewayv1.Install(scheme))
	utilruntime.Must(gatewayv1alpha2.Install(scheme))
	utilruntime.Must(gatewayv1beta1.Install(scheme))
	utilruntime.Must(cmv1.AddToScheme(scheme))
	utilruntime.Must(helmv2.AddToScheme(scheme))
	utilruntime.Must(cosiv1alpha1.AddToScheme(scheme))
	// +kubebuilder:scaffold:scheme
}

func main() {
	var metricsAddr string
	var enableLeaderElection bool
	var probeAddr string
	var secureMetrics bool
	var enableHTTP2 bool
	var disableTelemetry bool
	var telemetryEndpoint string
	var telemetryInterval string
	var quotaBufferPercent int64
	var seaweedfsMetricsEndpoint string
	var tlsOpts []func(*tls.Config)
	var gpuCfg gpu.Config
	var gpuResourceNames, gpuRawResourceNames string
	flag.StringVar(&metricsAddr, "metrics-bind-address", "0", "The address the metrics endpoint binds to. "+
		"Use :8443 for HTTPS or :8080 for HTTP, or leave as 0 to disable the metrics service.")
	flag.StringVar(&probeAddr, "health-probe-bind-address", ":8081", "The address the probe endpoint binds to.")
	flag.BoolVar(&enableLeaderElection, "leader-elect", false,
		"Enable leader election for controller manager. "+
			"Enabling this will ensure there is only one active controller manager.")
	flag.BoolVar(&secureMetrics, "metrics-secure", true,
		"If set, the metrics endpoint is served securely via HTTPS. Use --metrics-secure=false to use HTTP instead.")
	flag.BoolVar(&enableHTTP2, "enable-http2", false,
		"If set, HTTP/2 will be enabled for the metrics and webhook servers")
	flag.BoolVar(&disableTelemetry, "disable-telemetry", false,
		"Disable telemetry collection")
	flag.StringVar(&telemetryEndpoint, "telemetry-endpoint", "https://telemetry.cozystack.io",
		"Endpoint for sending telemetry data")
	flag.StringVar(&telemetryInterval, "telemetry-interval", "15m",
		"Interval between telemetry data collection (e.g. 15m, 1h)")
	flag.Int64Var(&quotaBufferPercent, "tenant-quota-buffer-percent", 0,
		"Temporary buffer (e.g. 130 = +30%) added to every hierarchical tenant quota pool so workloads already over a freshly-introduced quota keep running during rollout. 0 disables it.")
	flag.StringVar(&seaweedfsMetricsEndpoint, "seaweedfs-metrics-endpoint", "",
		"Base URL of a Prometheus-compatible query API to fetch SeaweedFS bucket size metrics from, e.g. https://vm.example.com/path/to/prometheus (/api/v1/query is appended). "+
			"Overrides discovery via the namespace.cozystack.io/monitoring label; use when SeaweedFS and the monitoring stack that scrapes it run in a separate cluster. "+
			"Basic auth may be embedded as userinfo (https://user:pass@host/...). Empty keeps label-based discovery.")
	flag.BoolVar(&gpuCfg.Enabled, "enable-gpu", false,
		"Enable the GPU features that need KubeVirt, such as GPUClaim placement. GPU finalizers and usage records run without it.")
	flag.StringVar(&gpuResourceNames, "gpu-resource-names", "",
		"Comma-separated GPU resource names besides those of GPUClasses and the node agents (gpu.resourceNames).")
	flag.StringVar(&gpuRawResourceNames, "gpu-raw-resource-names", string(gpu.RawNamesAllowed),
		"Whether VMs and pods may request GPU resource names directly: allowed or forbidden (gpu.rawResourceNames).")
	flag.DurationVar(&gpuCfg.LostCapacityGrace, "gpu-lost-capacity-grace", 5*time.Minute,
		"How long a GPU node stays NotReady, or serves fewer units of a GPU resource than its devices hold, before its claims move.")
	flag.DurationVar(&gpuCfg.AttachmentGrace, "gpu-attachment-grace", 2*time.Minute,
		"How long a GPUClaim attachment waits for its VMI to appear; must exceed the API server request timeout.")
	opts := zap.Options{
		Development: false,
	}
	opts.BindFlags(flag.CommandLine)
	flag.Parse()

	// Parse telemetry interval
	interval, err := time.ParseDuration(telemetryInterval)
	if err != nil {
		setupLog.Error(err, "invalid telemetry interval")
		os.Exit(1)
	}

	// Validate and normalize the SeaweedFS metrics endpoint override; the sizes
	// it serves feed billing, so a malformed URL must fail startup, not queries.
	if seaweedfsMetricsEndpoint != "" {
		seaweedfsMetricsEndpoint, err = controller.ParseMetricsEndpointURL(seaweedfsMetricsEndpoint)
		if err != nil {
			setupLog.Error(err, "invalid --seaweedfs-metrics-endpoint")
			os.Exit(1)
		}
	}

	gpuCfg.ResourceNames = model.ParseResourceNames(gpuResourceNames)
	gpuCfg.RawResourceNames = gpu.RawNamesPolicy(gpuRawResourceNames)
	if err := validateGPUConfig(&gpuCfg); err != nil {
		setupLog.Error(err, "invalid GPU flags")
		os.Exit(1)
	}

	// Configure telemetry
	telemetryConfig := telemetry.Config{
		Disabled: disableTelemetry,
		Endpoint: telemetryEndpoint,
		Interval: interval,
	}

	ctrl.SetLogger(zap.New(zap.UseFlagOptions(&opts)))

	// if the enable-http2 flag is false (the default), http/2 should be disabled
	// due to its vulnerabilities. More specifically, disabling http/2 will
	// prevent from being vulnerable to the HTTP/2 Stream Cancellation and
	// Rapid Reset CVEs. For more information see:
	// - https://github.com/advisories/GHSA-qppj-fm5r-hxr3
	// - https://github.com/advisories/GHSA-4374-p667-p6c8
	disableHTTP2 := func(c *tls.Config) {
		setupLog.Info("disabling http/2")
		c.NextProtos = []string{"http/1.1"}
	}

	if !enableHTTP2 {
		tlsOpts = append(tlsOpts, disableHTTP2)
	}

	webhookServer := webhook.NewServer(webhook.Options{
		TLSOpts: tlsOpts,
	})

	// Metrics endpoint is enabled in 'config/default/kustomization.yaml'. The Metrics options configure the server.
	// More info:
	// - https://pkg.go.dev/sigs.k8s.io/controller-runtime@v0.19.1/pkg/metrics/server
	// - https://book.kubebuilder.io/reference/metrics.html
	metricsServerOptions := metricsserver.Options{
		BindAddress:   metricsAddr,
		SecureServing: secureMetrics,
		TLSOpts:       tlsOpts,
	}

	if secureMetrics {
		// FilterProvider is used to protect the metrics endpoint with authn/authz.
		// These configurations ensure that only authorized users and service accounts
		// can access the metrics endpoint. The RBAC are configured in 'config/rbac/kustomization.yaml'. More info:
		// https://pkg.go.dev/sigs.k8s.io/controller-runtime@v0.19.1/pkg/metrics/filters#WithAuthenticationAndAuthorization
		metricsServerOptions.FilterProvider = filters.WithAuthenticationAndAuthorization

		// TODO(user): If CertDir, CertName, and KeyName are not specified, controller-runtime will automatically
		// generate self-signed certificates for the metrics server. While convenient for development and testing,
		// this setup is not recommended for production.
	}

	// Configure rate limiting for the Kubernetes client
	config := ctrl.GetConfigOrDie()
	config.QPS = 50.0  // Increased from default 5.0
	config.Burst = 100 // Increased from default 10

	mgr, err := ctrl.NewManager(config, ctrl.Options{
		Scheme:                 scheme,
		Metrics:                metricsServerOptions,
		WebhookServer:          webhookServer,
		HealthProbeBindAddress: probeAddr,
		LeaderElection:         enableLeaderElection,
		LeaderElectionID:       "19a0338c.cozystack.io",
		// Scope the shared Secret informer so the WildcardSecret reconciler's
		// cluster-wide Secret watch does not cache every Secret (and its key
		// material) in memory. Only managed wildcard replicas and the values
		// channel are cached; no other reconciler reads Secrets through THIS
		// (the manager's) typed Secret cache. See wildcardsecret.SecretCacheByObject.
		//
		// The CA-extraction reconciler does NOT use this cache at all — neither its
		// typed Secret informer nor a metadata one. A metadata informer for
		// v1/Secret would route to THIS same per-GVK cache and inherit its
		// wildcard-replica label selector, so its source watch would never fire for
		// an unlabelled CA source. It therefore owns a SEPARATE metadata-only cache
		// (caSecretCluster, below) and reads the one source it projects through the
		// uncached APIReader. So this manager-level Secret scoping is
		// wildcardsecret's alone.
		Cache: cache.Options{
			ByObject: map[client.Object]cache.ByObject{
				&corev1.Secret{}: wildcardsecret.SecretCacheByObject(),
				// The GPU settings ConfigMap is the only ConfigMap this binary
				// reads through the cache, so cache only that one.
				&corev1.ConfigMap{}: {
					Namespaces: map[string]cache.Config{gpu.SettingsNamespace: {
						FieldSelector: fields.OneTermEqualSelector("metadata.name", gpu.SettingsConfigMap),
					}},
				},
			},
		},
		// The program exits as soon as the manager stops, which is what makes
		// releasing the lease on cancel safe. Without it a rollout leaves the
		// launcher pods of starting GPU VMs gated for a whole lease duration.
		LeaderElectionReleaseOnCancel: true,
	})
	if err != nil {
		setupLog.Error(err, "unable to start manager")
		os.Exit(1)
	}

	if err = (&controller.WorkloadMonitorReconciler{
		Client:                   mgr.GetClient(),
		Scheme:                   mgr.GetScheme(),
		Recorder:                 mgr.GetEventRecorderFor("workloadmonitor-controller"),
		SeaweedfsMetricsEndpoint: seaweedfsMetricsEndpoint,
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "WorkloadMonitor")
		os.Exit(1)
	}

	if err = (&controller.WorkloadReconciler{
		Client: mgr.GetClient(),
		Scheme: mgr.GetScheme(),
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "WorkloadReconciler")
		os.Exit(1)
	}

	if err = (&controller.ApplicationDefinitionReconciler{
		Client: mgr.GetClient(),
		Scheme: mgr.GetScheme(),
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "ApplicationDefinitionReconciler")
		os.Exit(1)
	}

	if err = (&controller.ApplicationDefinitionHelmReconciler{
		Client:   mgr.GetClient(),
		Scheme:   mgr.GetScheme(),
		Recorder: mgr.GetEventRecorderFor("applicationdefinition-helm-reconciler"),
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "ApplicationDefinitionHelmReconciler")
		os.Exit(1)
	}

	if err = (&tenantgateway.Reconciler{
		Client: mgr.GetClient(),
		Scheme: mgr.GetScheme(),
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "TenantGateway")
		os.Exit(1)
	}

	if err = (&tenantquota.Reconciler{
		Client:        mgr.GetClient(),
		Scheme:        mgr.GetScheme(),
		Recorder:      mgr.GetEventRecorderFor("tenantquota-controller"),
		BufferPercent: quotaBufferPercent,
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "TenantQuota")
		os.Exit(1)
	}

	if err = (&tenantlogrouting.Reconciler{
		Client: mgr.GetClient(),
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "TenantLogRouting")
		os.Exit(1)
	}

	if err = (&wildcardsecret.Reconciler{
		Client:   mgr.GetClient(),
		Reader:   mgr.GetAPIReader(),
		Scheme:   mgr.GetScheme(),
		Recorder: mgr.GetEventRecorderFor("wildcardsecret"),
	}).SetupWithManager(mgr); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "WildcardSecret")
		os.Exit(1)
	}

	// The CA-extraction reconciler watches its source Secrets as metadata only, so
	// no private key ever enters a cache. It cannot watch them on the manager's
	// cache: that cache's Secret informer is label-scoped to WildcardSecret's
	// replicas (above), a metadata informer for the same v1/Secret GVK routes to
	// that same scoped informer and inherits the selector, and the watch would
	// then silently never fire for an unlabelled CA source — the projection would
	// appear only on the slow resync. So it owns a SEPARATE cache, metadata-only
	// and unscoped, holding Secret metadata stubs cluster-wide (no key material).
	// It is a cluster.Cluster so mgr.Add registers it in the manager's
	// cache-runnable group — started, waited-for-sync before the controllers run,
	// and stopped on shutdown, exactly like the shared cache.
	caSecretCluster, err := cluster.New(config, func(o *cluster.Options) {
		o.Scheme = scheme
	})
	if err != nil {
		setupLog.Error(err, "unable to build the CA source metadata cache")
		os.Exit(1)
	}
	if err = mgr.Add(caSecretCluster); err != nil {
		setupLog.Error(err, "unable to add the CA source metadata cache to the manager")
		os.Exit(1)
	}

	if err = (&cacert.Reconciler{
		Client:   mgr.GetClient(),
		Reader:   mgr.GetAPIReader(),
		Recorder: mgr.GetEventRecorderFor("cacert-controller"),
	}).SetupWithManager(mgr, caSecretCluster.GetCache()); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "CACert")
		os.Exit(1)
	}

	if err = gpucontroller.Setup(mgr, gpuCfg); err != nil {
		setupLog.Error(err, "unable to create controller", "controller", "GPU")
		os.Exit(1)
	}
	// Asking the manager for its webhook server starts it, and the server
	// needs a serving certificate, so registration waits for --enable-gpu.
	if gpuCfg.Enabled {
		if err = gpuwebhook.Register(mgr, gpuCfg); err != nil {
			setupLog.Error(err, "unable to register the GPU webhooks")
			os.Exit(1)
		}
	}

	// +kubebuilder:scaffold:builder

	if err := mgr.AddHealthzCheck("healthz", healthz.Ping); err != nil {
		setupLog.Error(err, "unable to set up health check")
		os.Exit(1)
	}
	// With GPU enabled, a replica is ready only once the GPU, Node,
	// ConfigMap and Namespace informers have synced. Every replica starts
	// them, so a replica that takes the lease starts the GPU controller on
	// synced caches. The GPU webhooks read none of them: they serve on every
	// replica while only the leader binds, so they read from the API, where
	// a lagging watch cannot make them admit what the cluster refuses.
	// Readiness does not wait for KubeVirt, whose kinds a fresh install can
	// serve only after this controller starts: packages that depend on
	// cozystack-controller wait for its readiness. The VMI and KubeVirt
	// informers start once their kinds are served.
	readyz := healthz.Ping
	if gpuCfg.Enabled {
		vmiGVK := schema.GroupVersionKind{Group: "kubevirt.io", Version: "v1", Kind: "VirtualMachineInstance"}
		kvGVK := vmiGVK.GroupVersion().WithKind("KubeVirt")
		vmi := &unstructured.Unstructured{}
		vmi.SetGroupVersionKind(vmiGVK)
		kv := &unstructured.Unstructured{}
		kv.SetGroupVersionKind(kvGVK)
		warmup := &gpucontroller.Warmup{
			Mapper:    mgr.GetRESTMapper(),
			Informers: mgr.GetCache(),
			Objects: []client.Object{&gpuv1alpha1.GPUClaim{}, &gpuv1alpha1.GPUClass{}, &gpuv1alpha1.GPUNode{},
				&corev1.Node{}, &corev1.ConfigMap{}, &corev1.Namespace{}},
			Kinds:           []schema.GroupVersionKind{vmiGVK, kvGVK},
			KubeVirtObjects: []client.Object{vmi, kv},
			Interval:        5 * time.Second,
			OnWait: func(k schema.GroupVersionKind, err error) {
				setupLog.Info("waiting for a KubeVirt kind the GPU webhooks read", "kind", k.String(), "reason", err.Error())
			},
		}
		if err := mgr.Add(warmup); err != nil {
			setupLog.Error(err, "unable to add the GPU informer warm-up")
			os.Exit(1)
		}
		readyz = warmup.Ready
	}
	if err := mgr.AddReadyzCheck("readyz", readyz); err != nil {
		setupLog.Error(err, "unable to set up ready check")
		os.Exit(1)
	}

	// Initialize telemetry collector
	collector, err := telemetry.NewCollector(mgr.GetClient(), &telemetryConfig, mgr.GetConfig())
	if err != nil {
		setupLog.V(1).Error(err, "unable to create telemetry collector, telemetry will be disabled")
	}

	if collector != nil {
		if err := mgr.Add(collector); err != nil {
			setupLog.Error(err, "unable to set up telemetry collector")
			setupLog.V(1).Error(err, "unable to set up telemetry collector, continuing without telemetry")
		}
	}

	setupLog.Info("starting manager")
	if err := mgr.Start(ctrl.SetupSignalHandler()); err != nil {
		setupLog.Error(err, "problem running manager")
		os.Exit(1)
	}
}

// validateGPUConfig defaults an empty raw-name policy to allowed and
// refuses non-positive grace periods, since a zero attachment grace would
// clear every attachment before its VMI appears.
func validateGPUConfig(cfg *gpu.Config) error {
	if cfg.RawResourceNames == "" {
		cfg.RawResourceNames = gpu.RawNamesAllowed
	}
	if cfg.LostCapacityGrace <= 0 || cfg.AttachmentGrace <= 0 {
		return errors.New("--gpu-lost-capacity-grace and --gpu-attachment-grace must be positive")
	}
	if cfg.RawResourceNames != gpu.RawNamesAllowed && cfg.RawResourceNames != gpu.RawNamesForbidden {
		return fmt.Errorf("--gpu-raw-resource-names is %q, want %s or %s", cfg.RawResourceNames, gpu.RawNamesAllowed, gpu.RawNamesForbidden)
	}
	return nil
}