Add support to sysctl node-exporter collector

Signed-off-by: Carlos Goncalves <[email protected]>

Author: cgoncalves

CHANGELOG.md

@@ -5,6 +5,7 @@
 - [#2302](https://github.com/openshift/cluster-monitoring-operator/issues/2302) Enable feature `extra-scrape-metrics` in Prometheus user-workload
 - [#2319](https://github.com/openshift/cluster-monitoring-operator/pull/2319) Allow read-only access to the Alertmanager API (use `monitoring-alertmanager-view`).
 - [#2078](https://github.com/openshift/cluster-monitoring-operator/pull/2078) Support exporting VPA metrics from KSM.
+- [#2339](https://github.com/openshift/cluster-monitoring-operator/pull/2339) Add support to sysctl node-exporter collector
 
 ## 4.15
 

Documentation/api.md

@@ -34,6 +34,7 @@ Configuring Cluster Monitoring is optional. If the config does not exist or is e
 * [NodeExporterCollectorNetClassConfig](#nodeexportercollectornetclassconfig)
 * [NodeExporterCollectorNetDevConfig](#nodeexportercollectornetdevconfig)
 * [NodeExporterCollectorProcessesConfig](#nodeexportercollectorprocessesconfig)
+* [NodeExporterCollectorSysctlConfig](#nodeexportercollectorsysctlconfig)
 * [NodeExporterCollectorSystemdConfig](#nodeexportercollectorsystemdconfig)
 * [NodeExporterCollectorTcpStatConfig](#nodeexportercollectortcpstatconfig)
 * [NodeExporterConfig](#nodeexporterconfig)
@@ -250,6 +251,7 @@ The `NodeExporterCollectorConfig` resource defines settings for individual colle
 | mountstats | [NodeExporterCollectorMountStatsConfig](#nodeexportercollectormountstatsconfig) | Defines the configuration of the `mountstats` collector, which collects statistics about NFS volume I/O activities. Disabled by default. |
 | ksmd | [NodeExporterCollectorKSMDConfig](#nodeexportercollectorksmdconfig) | Defines the configuration of the `ksmd` collector, which collects statistics from the kernel same-page merger daemon. Disabled by default. |
 | processes | [NodeExporterCollectorProcessesConfig](#nodeexportercollectorprocessesconfig) | Defines the configuration of the `processes` collector, which collects statistics from processes and threads running in the system. Disabled by default. |
+| sysctl | [NodeExporterCollectorSysctlConfig](#nodeexportercollectorsysctlconfig) | Defines the configuration of the `sysctl` collector, which collects sysctl metrics. Disabled by default. |
 | systemd | [NodeExporterCollectorSystemdConfig](#nodeexportercollectorsystemdconfig) | Defines the configuration of the `systemd` collector, which collects statistics on the systemd daemon and its managed services. Disabled by default. |
 
 [Back to TOC](#table-of-contents)
@@ -345,6 +347,23 @@ The `NodeExporterCollectorProcessesConfig` resource works as an on/off switch fo
 
 [Back to TOC](#table-of-contents)
 
+## NodeExporterCollectorSysctlConfig
+
+#### Description
+
+The `NodeExporterCollectorSysctlConfig` resource works as an on/off switch for the `sysctl` collector of the `node-exporter` agent. Caution! Exposing metrics like kernel.random.uuid can disrupt Prometheus, as it generates new data series with every scrape. Use this option judiciously! By default, the `sysctl` collector is disabled.
+
+
+<em>appears in: [NodeExporterCollectorConfig](#nodeexportercollectorconfig)</em>
+
+| Property | Type | Description |
+| -------- | ---- | ----------- |
+| enabled | bool | A Boolean flag that enables or disables the `sysctl` collector. |
+| includeSysctlMetrics | []string | A list of numeric sysctl values. Note that a sysctl can contain multiple values, for example: `net.ipv4.tcp_rmem = 4096\t131072\t6291456`. Using `includeSysctlMetrics: ['net.ipv4.tcp_rmem']` the collector will expose: `node_sysctl_net_ipv4_tcp_rmem{index=\"0\"} 4096`, `node_sysctl_net_ipv4_tcp_rmem{index=\"1\"} 131072`, `node_sysctl_net_ipv4_tcp_rmem{index=\"2\"} 6291456`. If the indexes have defined meaning like in this case, the values can be mapped to multiple metrics: `includeSysctlMetrics: ['net.ipv4.tcp_rmem:min,default,max']`. The collector will expose these metrics as such: `node_sysctl_net_ipv4_tcp_rmem_min 4096`, `node_sysctl_net_ipv4_tcp_rmem_default 131072`, `node_sysctl_net_ipv4_tcp_rmem_max 6291456`. |
+| includeInfoSysctlMetrics | []string | A list of string sysctl values. For example: `includeSysctlMetrics: ['kernel.core_pattern', 'kernel.seccomp.actions_avail = kill_process kill_thread']`. The collector will expose these metrics as such: `node_sysctl_info{name=\"kernel.core_pattern\", value=\"core\"} 1`, `node_sysctl_info{name=\"kernel.seccomp.actions_avail\", index=\"0\", value=\"kill_process\"} 1`, `node_sysctl_info{name=\"kernel.seccomp.actions_avail\", index=\"1\", value=\"kill_thread\"} 1`, ... |
+
+[Back to TOC](#table-of-contents)
+
 ## NodeExporterCollectorSystemdConfig
 
 #### Description

Documentation/openshiftdocs/index.adoc

@@ -54,6 +54,7 @@ The configuration file itself is always defined under the `config.yaml` key in t
 * link:modules/nodeexportercollectornetclassconfig.adoc[NodeExporterCollectorNetClassConfig]
 * link:modules/nodeexportercollectornetdevconfig.adoc[NodeExporterCollectorNetDevConfig]
 * link:modules/nodeexportercollectorprocessesconfig.adoc[NodeExporterCollectorProcessesConfig]
+* link:modules/nodeexportercollectorsysctlconfig.adoc[NodeExporterCollectorSysctlConfig]
 * link:modules/nodeexportercollectorsystemdconfig.adoc[NodeExporterCollectorSystemdConfig]
 * link:modules/nodeexportercollectortcpstatconfig.adoc[NodeExporterCollectorTcpStatConfig]
 * link:modules/nodeexporterconfig.adoc[NodeExporterConfig]

Documentation/openshiftdocs/modules/nodeexportercollectorconfig.adoc

@@ -34,6 +34,8 @@ Appears in: link:nodeexporterconfig.adoc[NodeExporterConfig]
 
 |processes|link:nodeexportercollectorprocessesconfig.adoc[NodeExporterCollectorProcessesConfig]|Defines the configuration of the `processes` collector, which collects statistics from processes and threads running in the system. Disabled by default.
 
+|sysctl|link:nodeexportercollectorsysctlconfig.adoc[NodeExporterCollectorSysctlConfig]|Defines the configuration of the `sysctl` collector, which collects sysctl metrics. Disabled by default.
+
 |systemd|link:nodeexportercollectorsystemdconfig.adoc[NodeExporterCollectorSystemdConfig]|Defines the configuration of the `systemd` collector, which collects statistics on the systemd daemon and its managed services. Disabled by default.
 
 |===

Documentation/openshiftdocs/modules/nodeexportercollectorsysctlconfig.adoc

@@ -0,0 +1,29 @@
+// DO NOT EDIT THE CONTENT IN THIS FILE. It is automatically generated from the 
+	// source code for the Cluster Monitoring Operator. Any changes made to this 
+	// file will be overwritten when the content is re-generated. If you wish to 
+	// make edits, read the docgen utility instructions in the source code for the 
+	// CMO.
+	:_content-type: ASSEMBLY
+
+== NodeExporterCollectorSysctlConfig
+
+=== Description
+
+The `NodeExporterCollectorSysctlConfig` resource works as an on/off switch for the `sysctl` collector of the `node-exporter` agent. Caution! Exposing metrics like kernel.random.uuid can disrupt Prometheus, as it generates new data series with every scrape. Use this option judiciously! By default, the `sysctl` collector is disabled.
+
+
+
+Appears in: link:nodeexportercollectorconfig.adoc[NodeExporterCollectorConfig]
+
+[options="header"]
+|===
+| Property | Type | Description 
+|enabled|bool|A Boolean flag that enables or disables the `sysctl` collector.
+
+|includeSysctlMetrics|[]string|A list of numeric sysctl values. Note that a sysctl can contain multiple values, for example: `net.ipv4.tcp_rmem = 4096\t131072\t6291456`. Using `includeSysctlMetrics: ['net.ipv4.tcp_rmem']` the collector will expose: `node_sysctl_net_ipv4_tcp_rmem{index=\"0\"} 4096`, `node_sysctl_net_ipv4_tcp_rmem{index=\"1\"} 131072`, `node_sysctl_net_ipv4_tcp_rmem{index=\"2\"} 6291456`. If the indexes have defined meaning like in this case, the values can be mapped to multiple metrics: `includeSysctlMetrics: ['net.ipv4.tcp_rmem:min,default,max']`. The collector will expose these metrics as such: `node_sysctl_net_ipv4_tcp_rmem_min 4096`, `node_sysctl_net_ipv4_tcp_rmem_default 131072`, `node_sysctl_net_ipv4_tcp_rmem_max 6291456`.
+
+|includeInfoSysctlMetrics|[]string|A list of string sysctl values. For example: `includeSysctlMetrics: ['kernel.core_pattern', 'kernel.seccomp.actions_avail = kill_process kill_thread']`. The collector will expose these metrics as such: `node_sysctl_info{name=\"kernel.core_pattern\", value=\"core\"} 1`, `node_sysctl_info{name=\"kernel.seccomp.actions_avail\", index=\"0\", value=\"kill_process\"} 1`, `node_sysctl_info{name=\"kernel.seccomp.actions_avail\", index=\"1\", value=\"kill_thread\"} 1`, ...
+
+|===
+
+link:../index.adoc[Back to TOC]

pkg/manifests/config.go

@@ -217,6 +217,11 @@ func defaultClusterMonitoringConfiguration() ClusterMonitoringConfiguration {
 				Systemd: NodeExporterCollectorSystemdConfig{
 					Enabled: false,
 				},
+				Sysctl: NodeExporterCollectorSysctlConfig{
+					Enabled:                  false,
+					IncludeSysctlMetrics:     []string{},
+					IncludeInfoSysctlMetrics: []string{},
+				},
 			},
 		},
 	}

pkg/manifests/manifests.go

@@ -876,6 +876,25 @@ func (f *Factory) updateNodeExporterArgs(args []string) ([]string, error) {
 		args = setArg(args, "--no-collector.tcpstat", "")
 	}
 
+	if f.config.ClusterMonitoringConfiguration.NodeExporterConfig.Collectors.Sysctl.Enabled {
+		includeSysctlMetrics := f.config.ClusterMonitoringConfiguration.NodeExporterConfig.Collectors.Sysctl.IncludeSysctlMetrics
+		includeInfoSysctlMetrics := f.config.ClusterMonitoringConfiguration.NodeExporterConfig.Collectors.Sysctl.IncludeInfoSysctlMetrics
+
+		args = setArg(args, "--collector.sysctl", "")
+
+		sysctlSet := uniqueSet(includeSysctlMetrics)
+		for _, sysctl := range sysctlSet {
+			args = append(args, fmt.Sprintf("--collector.sysctl.include=%s", sysctl))
+		}
+
+		sysctlSet = uniqueSet(includeInfoSysctlMetrics)
+		for _, sysctl := range sysctlSet {
+			args = append(args, fmt.Sprintf("--collector.sysctl.include-info=%s", sysctl))
+		}
+	} else {
+		args = setArg(args, "--no-collector.sysctl", "")
+	}
+
 	var excludedDevices string
 	if f.config.ClusterMonitoringConfiguration.NodeExporterConfig.Collectors.NetDev.Enabled ||
 		f.config.ClusterMonitoringConfiguration.NodeExporterConfig.Collectors.NetClass.Enabled {
@@ -2488,6 +2507,18 @@ func setArg(args []string, argName string, argValue string) []string {
 	return args
 }
 
+func uniqueSet(input []string) []string {
+	uniqueMap := make(map[string]struct{})
+	var unique []string
+	for _, str := range input {
+		if _, ok := uniqueMap[str]; !ok {
+			uniqueMap[str] = struct{}{}
+			unique = append(unique, str)
+		}
+	}
+	return unique
+}
+
 func (f *Factory) PrometheusRuleValidatingWebhook() (*admissionv1.ValidatingWebhookConfiguration, error) {
 	return f.NewValidatingWebhook(f.assets.MustNewAssetSlice(AdmissionWebhookRuleValidatingWebhook))
 }

pkg/manifests/manifests_test.go

@@ -3480,6 +3480,7 @@ func TestNodeExporterCollectorSettings(t *testing.T) {
 			name:   "default config",
 			config: "",
 			argsPresent: []string{"--no-collector.cpufreq",
+				"--no-collector.sysctl",
 				"--no-collector.tcpstat",
 				"--collector.netdev",
 				"--collector.netclass",
@@ -3492,6 +3493,7 @@ func TestNodeExporterCollectorSettings(t *testing.T) {
 				"--no-collector.systemd",
 			},
 			argsAbsent: []string{"--collector.cpufreq",
+				"--collector.sysctl",
 				"--collector.tcpstat",
 				"--no-collector.netdev",
 				"--no-collector.netclass",
@@ -3657,6 +3659,38 @@ nodeExporter:
 				"--collector.systemd.unit-include=^(network.+|nss.+)$"},
 			argsAbsent: []string{"--no-collector.systemd"},
 		},
+		{
+			name: "disable sysctl collector",
+			config: `
+nodeExporter:
+  collectors:
+    sysctl:
+      enabled: false
+`,
+			argsPresent: []string{"--no-collector.sysctl"},
+			argsAbsent:  []string{"--collector.sysctl"},
+		},
+		{
+			name: "enable sysctl collector",
+			config: `
+nodeExporter:
+  collectors:
+    sysctl:
+      enabled: true
+      includeSysctlMetrics:
+        - net.ipv4.tcp_rmem:min,default,max
+        - net.ipv4.tcp_mem
+      includeInfoSysctlMetrics:
+        - kernel.core_pattern
+        - kernel.seccomp.actions_avail
+`,
+			argsPresent: []string{"--collector.sysctl",
+				"--collector.sysctl.include=net.ipv4.tcp_rmem:min,default,max",
+				"--collector.sysctl.include=net.ipv4.tcp_mem",
+				"--collector.sysctl.include-info=kernel.core_pattern",
+				"--collector.sysctl.include-info=kernel.seccomp.actions_avail"},
+			argsAbsent: []string{"--no-collector.sysctl"},
+		},
 	}
 
 	for _, test := range tests {

pkg/manifests/types.go

@@ -351,6 +351,9 @@ type NodeExporterCollectorConfig struct {
 	// Defines the configuration of the `processes` collector, which collects statistics from processes and threads running in the system.
 	// Disabled by default.
 	Processes NodeExporterCollectorProcessesConfig `json:"processes,omitempty"`
+	// Defines the configuration of the `sysctl` collector, which collects sysctl metrics.
+	// Disabled by default.
+	Sysctl NodeExporterCollectorSysctlConfig `json:"sysctl,omitempty"`
 	// Defines the configuration of the `systemd` collector, which collects statistics on the systemd daemon and its managed services.
 	// Disabled by default.
 	Systemd NodeExporterCollectorSystemdConfig `json:"systemd,omitempty"`
@@ -376,6 +379,38 @@ type NodeExporterCollectorTcpStatConfig struct {
 	Enabled bool `json:"enabled,omitempty"`
 }
 
+// The `NodeExporterCollectorSysctlConfig` resource works as an on/off switch for
+// the `sysctl` collector of the `node-exporter` agent.
+// Caution! Exposing metrics like kernel.random.uuid can disrupt Prometheus, as it generates new data series with every scrape. Use this option judiciously!
+// By default, the `sysctl` collector is disabled.
+type NodeExporterCollectorSysctlConfig struct {
+	// A Boolean flag that enables or disables the `sysctl` collector.
+	Enabled bool `json:"enabled,omitempty"`
+	// A list of numeric sysctl values.
+	// Note that a sysctl can contain multiple values, for example:
+	// `net.ipv4.tcp_rmem = 4096	131072	6291456`.
+	// Using `includeSysctlMetrics: ['net.ipv4.tcp_rmem']` the collector will expose:
+	// `node_sysctl_net_ipv4_tcp_rmem{index="0"} 4096`,
+	// `node_sysctl_net_ipv4_tcp_rmem{index="1"} 131072`,
+	// `node_sysctl_net_ipv4_tcp_rmem{index="2"} 6291456`.
+	// If the indexes have defined meaning like in this case, the values can be mapped to multiple metrics:
+	// `includeSysctlMetrics: ['net.ipv4.tcp_rmem:min,default,max']`.
+	// The collector will expose these metrics as such:
+	// `node_sysctl_net_ipv4_tcp_rmem_min 4096`,
+	// `node_sysctl_net_ipv4_tcp_rmem_default 131072`,
+	// `node_sysctl_net_ipv4_tcp_rmem_max 6291456`.
+	IncludeSysctlMetrics []string `json:"includeSysctlMetrics,omitempty"`
+	// A list of string sysctl values.
+	// For example:
+	// `includeSysctlMetrics: ['kernel.core_pattern', 'kernel.seccomp.actions_avail = kill_process kill_thread']`.
+	// The collector will expose these metrics as such:
+	// `node_sysctl_info{name="kernel.core_pattern", value="core"} 1`,
+	// `node_sysctl_info{name="kernel.seccomp.actions_avail", index="0", value="kill_process"} 1`,
+	// `node_sysctl_info{name="kernel.seccomp.actions_avail", index="1", value="kill_thread"} 1`,
+	// ...
+	IncludeInfoSysctlMetrics []string `json:"includeInfoSysctlMetrics,omitempty"`
+}
+
 // The `NodeExporterCollectorNetDevConfig` resource works as an on/off switch for
 // the `netdev` collector of the `node-exporter` agent.
 // By default, the `netdev` collector is enabled.