Merge branch 'main' into bertinmeshita/sc-18608/increase-cli-low-reso…

…urces

.goreleaser.yaml

@@ -11,6 +11,7 @@ builds:
     goos:
       - linux
       - darwin
+      - windows
     goarch:
       - amd64
       - arm64

CHANGELOG.md

@@ -28,6 +28,70 @@ We use the following categories for changes:
 
 ### Security
 
+## [0.22.6] 2024-10-21
+
+### Added
+
+### Changed
+
+### Fixed
+
+- Low resource preset should take precedence over kernel 5.11 overrides [#sc-20056]
+
+### Removed
+
+### Deprecated
+
+### Security
+
+## [0.22.5] 2024-10-13
+
+### Added
+
+### Changed
+
+- Update low resources presets to include new components [#sc-20036]
+
+### Fixed
+
+### Removed
+
+### Deprecated
+
+### Security
+
+## [0.22.4] 2024-10-07
+
+### Added
+
+### Changed
+
+### Fixed
+
+- Waiting for sensor pods instead of alligator pods [#sc-19833]
+
+### Removed
+
+### Deprecated
+
+### Security
+
+## [0.22.3] 2024-09-17
+
+### Added
+
+- Support groundcover CLI in windows [#sc-19269]
+
+### Changed
+
+### Fixed
+
+### Removed
+
+### Deprecated
+
+### Security
+
 ## [0.22.2] 2024-07-04
 
 ### Added

cmd/deploy.go

@@ -6,7 +6,6 @@ import (
 	"strconv"
 	"time"
 
-	"github.com/blang/semver/v4"
 	"github.com/getsentry/sentry-go"
 	"github.com/imdario/mergo"
 	"github.com/pkg/errors"
@@ -46,7 +45,6 @@ const (
 	HELM_REPO_URL                     = "https://helm.groundcover.com"
 	CLUSTER_URL_FORMAT                = "%s/?clusterId=%s&viewType=Overview"
 	QUAY_REGISTRY_PRESET_PATH         = "presets/quay.yaml"
-	AGENT_KERNEL_5_11_PRESET_PATH     = "presets/agent/kernel-5-11.yaml"
 	CUSTOM_METRICS_PRESET_PATH        = "presets/backend/custom-metrics.yaml"
 	KUBE_STATE_METRICS_PRESET_PATH    = "presets/backend/kube-state-metrics.yaml"
 	STORAGE_CLASS_TEMPLATE_PATH       = "templates/backend/storage-class.yaml"
@@ -471,7 +469,7 @@ func validateInstall(ctx context.Context, kubeClient *k8s.Client, releaseName, n
 	}
 
 	if agentEnabled {
-		if err = waitForAlligators(ctx, kubeClient, namespace, appVersion, deployableNodesCount, sentryHelmContext); err != nil {
+		if err = waitForSensors(ctx, kubeClient, namespace, appVersion, deployableNodesCount, sentryHelmContext); err != nil {
 			return err
 		}
 	}
@@ -607,7 +605,7 @@ func generateChartValues(chartValues map[string]interface{}, apiKey, installatio
 			helm.BACKEND_LOW_RESOURCES_PATH,
 		}
 	} else {
-		agentPresetPath := helm.GetAgentResourcePresetPath(allocatableResources)
+		agentPresetPath := helm.GetAgentResourcePresetPath(allocatableResources, nodesReport.MaximalKernelVersion())
 		if agentPresetPath != helm.DEFAULT_PRESET {
 			overridePaths = append(overridePaths, agentPresetPath)
 		}
@@ -632,10 +630,6 @@ func generateChartValues(chartValues map[string]interface{}, apiKey, installatio
 		overridePaths = append(overridePaths, KUBE_STATE_METRICS_PRESET_PATH)
 	}
 
-	if semver.MustParseRange(">=5.11.0")(nodesReport.MaximalKernelVersion()) {
-		overridePaths = append(overridePaths, AGENT_KERNEL_5_11_PRESET_PATH)
-	}
-
 	if len(overridePaths) > 0 {
 		sentryHelmContext.ResourcesPresets = overridePaths
 		sentryHelmContext.SetOnCurrentScope()

cmd/status.go

@@ -28,18 +28,18 @@ const (
 	PVC_POLLING_RETRIES  = 40
 	PVC_POLLING_TIMEOUT  = time.Minute * 10
 
-	ALLIGATORS_POLLING_INTERVAL = time.Second * 15
-	ALLIGATORS_POLLING_RETRIES  = 28
-	ALLIGATORS_POLLING_TIMEOUT  = time.Minute * 7
+	SENSORS_POLLING_INTERVAL = time.Second * 15
+	SENSORS_POLLING_RETRIES  = 28
+	SENSORS_POLLING_TIMEOUT  = time.Minute * 7
 
-	ALLIGATOR_LABEL_SELECTOR = "app=alligator"
-	BACKEND_LABEL_SELECTOR   = "app!=alligator"
-	PORTAL_LABEL_SELECTOR    = "app=portal"
-	RUNNING_FIELD_SELECTOR   = "status.phase=Running"
+	SENSOR_LABEL_SELECTOR  = "app=sensor"
+	BACKEND_LABEL_SELECTOR = "app!=sensor"
+	PORTAL_LABEL_SELECTOR  = "app=portal"
+	RUNNING_FIELD_SELECTOR = "status.phase=Running"
 
 	WAIT_FOR_PORTAL_FORMAT      = "Waiting until cluster establish connectivity"
 	WAIT_FOR_PVCS_FORMAT        = "Waiting until all PVCs are bound (%d/%d PVCs)"
-	WAIT_FOR_ALLIGATORS_FORMAT  = "Waiting until all nodes are monitored (%d/%d Nodes)"
+	WAIT_FOR_SENSORS_FORMAT     = "Waiting until all nodes are monitored (%d/%d Nodes)"
 	TIMEOUT_INSTALLATION_FORMAT = "Installation takes longer than expected, you can check the status using \"kubectl get pods -n %s\""
 
 	PVCS_VALIDATION_EVENT_NAME   = "pvcs_validation"
@@ -117,7 +117,7 @@ var StatusCmd = &cobra.Command{
 		}
 		nodesCount := len(nodeList.Items)
 
-		if err = waitForAlligators(ctx, kubeClient, namespace, chart.AppVersion(), nodesCount, sentryHelmContext); err != nil {
+		if err = waitForSensors(ctx, kubeClient, namespace, chart.AppVersion(), nodesCount, sentryHelmContext); err != nil {
 			return err
 		}
 
@@ -178,7 +178,7 @@ func waitForPortal(ctx context.Context, kubeClient *k8s.Client, namespace, appVe
 	return err
 }
 
-func waitForAlligators(ctx context.Context, kubeClient *k8s.Client, namespace, appVersion string, expectedAlligatorsCount int, sentryHelmContext *sentry_utils.HelmContext) error {
+func waitForSensors(ctx context.Context, kubeClient *k8s.Client, namespace, appVersion string, expectedSensorsCount int, sentryHelmContext *sentry_utils.HelmContext) error {
 	var err error
 
 	event := segment.NewEvent(AGENTS_VALIDATION_EVENT_NAME)
@@ -187,43 +187,43 @@ func waitForAlligators(ctx context.Context, kubeClient *k8s.Client, namespace, a
 		event.StatusByError(err)
 	}()
 
-	spinner := ui.GlobalWriter.NewSpinner(fmt.Sprintf(WAIT_FOR_ALLIGATORS_FORMAT, 0, expectedAlligatorsCount))
-	spinner.SetStopMessage(fmt.Sprintf("All nodes are monitored (%d/%d Nodes)", expectedAlligatorsCount, expectedAlligatorsCount))
+	spinner := ui.GlobalWriter.NewSpinner(fmt.Sprintf(WAIT_FOR_SENSORS_FORMAT, 0, expectedSensorsCount))
+	spinner.SetStopMessage(fmt.Sprintf("All nodes are monitored (%d/%d Nodes)", expectedSensorsCount, expectedSensorsCount))
 	spinner.SetStopFailMessage(fmt.Sprintf(TIMEOUT_INSTALLATION_FORMAT, namespace))
 
 	spinner.Start()
 	defer spinner.WriteStop()
 
-	runningAlligators := 0
+	runningSensors := 0
 
-	isAlligatorRunningFunc := func() error {
+	isSensorRunningFunc := func() error {
 		var err error
 
-		if runningAlligators, err = getRunningAlligators(ctx, kubeClient, appVersion, namespace); err != nil {
+		if runningSensors, err = getRunningSensors(ctx, kubeClient, appVersion, namespace); err != nil {
 			return err
 		}
 
-		spinner.WriteMessage(fmt.Sprintf(WAIT_FOR_ALLIGATORS_FORMAT, runningAlligators, expectedAlligatorsCount))
+		spinner.WriteMessage(fmt.Sprintf(WAIT_FOR_SENSORS_FORMAT, runningSensors, expectedSensorsCount))
 
-		if runningAlligators >= expectedAlligatorsCount {
+		if runningSensors >= expectedSensorsCount {
 			return nil
 		}
 
-		err = errors.New("not all expected alligators are running")
+		err = errors.New("not all expected sensors are running")
 		return ui.RetryableError(err)
 	}
 
-	err = spinner.Poll(ctx, isAlligatorRunningFunc, ALLIGATORS_POLLING_INTERVAL, ALLIGATORS_POLLING_TIMEOUT, ALLIGATORS_POLLING_RETRIES)
+	err = spinner.Poll(ctx, isSensorRunningFunc, SENSORS_POLLING_INTERVAL, SENSORS_POLLING_TIMEOUT, SENSORS_POLLING_RETRIES)
 
-	runningAlligatorsStr := fmt.Sprintf("%d/%d", runningAlligators, expectedAlligatorsCount)
-	sentryHelmContext.RunningAlligators = runningAlligatorsStr
-	sentry_utils.SetTagOnCurrentScope(sentry_utils.EXPECTED_NODES_COUNT_TAG, fmt.Sprintf("%d", expectedAlligatorsCount))
-	sentry_utils.SetTagOnCurrentScope(sentry_utils.RUNNING_ALLIGATORS_TAG, runningAlligatorsStr)
+	runningSensorsStr := fmt.Sprintf("%d/%d", runningSensors, expectedSensorsCount)
+	sentryHelmContext.RunningSensors = runningSensorsStr
+	sentry_utils.SetTagOnCurrentScope(sentry_utils.EXPECTED_NODES_COUNT_TAG, fmt.Sprintf("%d", expectedSensorsCount))
+	sentry_utils.SetTagOnCurrentScope(sentry_utils.RUNNING_SENSORS_TAG, runningSensorsStr)
 
 	sentryHelmContext.SetOnCurrentScope()
 	event.
-		Set("alligatorsCount", expectedAlligatorsCount).
-		Set("runningAlligatorsCount", runningAlligators)
+		Set("sensorsCount", expectedSensorsCount).
+		Set("runningSensorsCount", runningSensors)
 
 	if err == nil {
 		return nil
@@ -232,9 +232,9 @@ func waitForAlligators(ctx context.Context, kubeClient *k8s.Client, namespace, a
 	defer spinner.WriteStopFail()
 
 	if errors.Is(err, ui.ErrSpinnerTimeout) {
-		if runningAlligators > 0 {
+		if runningSensors > 0 {
 			spinner.SetWarningSign()
-			spinner.SetStopFailMessage(fmt.Sprintf("groundcover managed to provision %d/%d nodes", runningAlligators, expectedAlligatorsCount))
+			spinner.SetStopFailMessage(fmt.Sprintf("groundcover managed to provision %d/%d nodes", runningSensors, expectedSensorsCount))
 		}
 
 		return ErrExecutionPartialSuccess
@@ -243,27 +243,27 @@ func waitForAlligators(ctx context.Context, kubeClient *k8s.Client, namespace, a
 	return err
 }
 
-func getRunningAlligators(ctx context.Context, kubeClient *k8s.Client, appVersion string, namespace string) (int, error) {
+func getRunningSensors(ctx context.Context, kubeClient *k8s.Client, appVersion string, namespace string) (int, error) {
 	podClient := kubeClient.CoreV1().Pods(namespace)
 	listOptions := metav1.ListOptions{
-		LabelSelector: ALLIGATOR_LABEL_SELECTOR,
+		LabelSelector: SENSOR_LABEL_SELECTOR,
 		FieldSelector: RUNNING_FIELD_SELECTOR,
 	}
 
-	runningAlligators := 0
+	runningSensors := 0
 
 	podList, err := podClient.List(ctx, listOptions)
 	if err != nil {
-		return runningAlligators, err
+		return runningSensors, err
 	}
 
 	for _, pod := range podList.Items {
 		if pod.Annotations["groundcover_version"] == appVersion {
-			runningAlligators++
+			runningSensors++
 		}
 	}
 
-	return runningAlligators, nil
+	return runningSensors, nil
 }
 
 func reportPodsStatus(ctx context.Context, kubeClient *k8s.Client, namespace string, sentryHelmContext *sentry_utils.HelmContext) {
@@ -272,7 +272,7 @@ func reportPodsStatus(ctx context.Context, kubeClient *k8s.Client, namespace str
 		return
 	}
 
-	agentPodsStatus, err := listPodsStatuses(ctx, kubeClient, namespace, metav1.ListOptions{LabelSelector: ALLIGATOR_LABEL_SELECTOR})
+	agentPodsStatus, err := listPodsStatuses(ctx, kubeClient, namespace, metav1.ListOptions{LabelSelector: SENSOR_LABEL_SELECTOR})
 	if err != nil {
 		return
 	}

pkg/helm/presets/agent/kernel-5-11.yaml

@@ -1,6 +1,6 @@
 ---
 agent:
-  alligator:
+  sensor:
     resources:
       requests:
         memory: 1Gi

pkg/helm/presets/agent/low-resources.yaml

@@ -1,6 +1,6 @@
 ---
 agent:
-  alligator:
+  sensor:
     resources:
       requests:
         cpu: 100m

pkg/helm/presets/backend/low-resources.yaml

@@ -73,7 +73,6 @@ monitors-manager:
     limits:
       memory: 150Mi
 
-
 backend:
   postgresql:
     primary:
@@ -83,4 +82,18 @@ backend:
           memory: 40Mi
         limits:
           memory: 60Mi
+  keep:
+    backend:
+      resources:
+        requests:
+          cpu: 100m
+          memory: 128Mi
 
+vector:
+  replicas: 1
+  resources:
+    requests:
+      cpu: 50m
+      memory: 256Mi
+    limits:
+      memory: 1024Mi

pkg/helm/tune.go

@@ -3,6 +3,7 @@ package helm
 import (
 	"embed"
 
+	"github.com/blang/semver/v4"
 	"groundcover.com/pkg/k8s"
 	"k8s.io/apimachinery/pkg/api/resource"
 )
@@ -17,6 +18,13 @@ const (
 	AGENT_DEFAULT_MEMORY_THRESHOLD = "1024Mi"
 	AGENT_LOW_RESOURCES_PATH       = "presets/agent/low-resources.yaml"
 
+	// Starting from Linux kernel version 5.11, eBPF maps are accounted for in the memory cgroup
+	// of the process that created them. For this reason we need to increase the memory limit for
+	// the agent.
+	// https://github.com/cilium/ebpf/blob/v0.16.0/docs/ebpf/concepts/rlimit.md#resource-limits
+	AGENT_KERNEL_5_11_PRESET_PATH = "presets/agent/kernel-5-11.yaml"
+	KERNEL_5_11_SEMVER_EXPRESSION = ">=5.11.0"
+
 	EMPTYDIR_STORAGE_PATH = "presets/backend/emptydir-storage.yaml"
 
 	BACKEND_DEFAULT_TOTAL_CPU_THRESHOLD    = "12000m"
@@ -39,7 +47,7 @@ type AllocatableResources struct {
 	NodeCount   int
 }
 
-func GetAgentResourcePresetPath(allocatableResources *AllocatableResources) string {
+func GetAgentResourcePresetPath(allocatableResources *AllocatableResources, maxKernelVersion semver.Version) string {
 	defaultCpuThreshold := resource.MustParse(AGENT_DEFAULT_CPU_THRESHOLD)
 	defaultMemoryThreshold := resource.MustParse(AGENT_DEFAULT_MEMORY_THRESHOLD)
 
@@ -50,6 +58,10 @@ func GetAgentResourcePresetPath(allocatableResources *AllocatableResources) stri
 		return AGENT_LOW_RESOURCES_PATH
 	}
 
+	if semver.MustParseRange(KERNEL_5_11_SEMVER_EXPRESSION)(maxKernelVersion) {
+		return AGENT_KERNEL_5_11_PRESET_PATH
+	}
+
 	return DEFAULT_PRESET
 }