[metricbeat][system/process, system/process_summary]: mark module as …

…healthy if metrics are partially filled (#40565) * chore: mark module as healthy if metrics are partially filled * chore: mark module as healthy if metrics are partially filled * fix: use errors.As * fix: fix lint * Update metricbeat/mb/event.go Co-authored-by: Mauri de Souza Meneguzzo <mauri870@gmail.com> * fix: changelog --------- Co-authored-by: Mauri de Souza Meneguzzo <mauri870@gmail.com>
elastic · Aug 23, 2024 · 1f033c9 · 1f033c9
1 parent 94ca509
commit 1f033c9
Show file tree

Hide file tree

Showing 5 changed files with 35 additions and 2 deletions.
diff --git a/CHANGELOG.next.asciidoc b/CHANGELOG.next.asciidoc
@@ -56,6 +56,7 @@ https://github.com/elastic/beats/compare/v8.8.1\...main[Check the HEAD diff]
 - Add support for Kibana status metricset in v8 format {pull}40275[40275]
 - Add new metrics for the vSphere Datastore metricset. {pull}40441[40441]
 - Update metrics for the vSphere Host metricset. {pull}40429[40429]
+- Mark system process metricsets as running if metrics are partially available {pull}40565[40565]
 - Added back `elasticsearch.node.stats.jvm.mem.pools.*` to the `node_stats` metricset {pull}40571[40571]
 
 *Osquerybeat*

diff --git a/metricbeat/mb/event.go b/metricbeat/mb/event.go
@@ -214,3 +214,18 @@ func tryToMapStr(v interface{}) (mapstr.M, bool) {
 		return nil, false
 	}
 }
+
+// PartialMetricsError indicates that metrics are only partially filled.
+// This will be removed once we fix the underlying problem.
+// See https://github.com/elastic/beats/issues/40542 for more details.
+type PartialMetricsError struct {
+	Err error
+}
+
+func (p PartialMetricsError) Error() string {
+	return p.Err.Error()
+}
+
+func (p PartialMetricsError) Unwrap() error {
+	return p.Err
+}
diff --git a/metricbeat/mb/module/wrapper.go b/metricbeat/mb/module/wrapper.go
@@ -19,6 +19,7 @@ package module
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"math/rand"
 	"sync"
@@ -255,7 +256,13 @@ func (msw *metricSetWrapper) fetch(ctx context.Context, reporter reporter) {
 		err := fetcher.Fetch(reporter.V2())
 		if err != nil {
 			reporter.V2().Error(err)
-			msw.module.UpdateStatus(status.Degraded, fmt.Sprintf("Error fetching data for metricset %s.%s: %v", msw.module.Name(), msw.MetricSet.Name(), err))
+			if errors.As(err, &mb.PartialMetricsError{}) {
+				// mark module as running if metrics are partially available and display the error message
+				msw.module.UpdateStatus(status.Running, fmt.Sprintf("Error fetching data for metricset %s.%s: %v", msw.module.Name(), msw.MetricSet.Name(), err))
+			} else {
+				// mark it as degraded for any other issue encountered
+				msw.module.UpdateStatus(status.Degraded, fmt.Sprintf("Error fetching data for metricset %s.%s: %v", msw.module.Name(), msw.MetricSet.Name(), err))
+			}
 			logp.Err("Error fetching data for metricset %s.%s: %s", msw.module.Name(), msw.Name(), err)
 		} else {
 			msw.module.UpdateStatus(status.Running, "")
@@ -265,7 +272,13 @@ func (msw *metricSetWrapper) fetch(ctx context.Context, reporter reporter) {
 		err := fetcher.Fetch(ctx, reporter.V2())
 		if err != nil {
 			reporter.V2().Error(err)
-			msw.module.UpdateStatus(status.Degraded, fmt.Sprintf("Error fetching data for metricset %s.%s: %v", msw.module.Name(), msw.MetricSet.Name(), err))
+			if errors.As(err, &mb.PartialMetricsError{}) {
+				// mark module as running if metrics are partially available and display the error message
+				msw.module.UpdateStatus(status.Running, fmt.Sprintf("Error fetching data for metricset %s.%s: %v", msw.module.Name(), msw.MetricSet.Name(), err))
+			} else {
+				// mark it as degraded for any other issue encountered
+				msw.module.UpdateStatus(status.Degraded, fmt.Sprintf("Error fetching data for metricset %s.%s: %v", msw.module.Name(), msw.MetricSet.Name(), err))
+			}
 			logp.Err("Error fetching data for metricset %s.%s: %s", msw.module.Name(), msw.Name(), err)
 		} else {
 			msw.module.UpdateStatus(status.Running, "")

diff --git a/metricbeat/module/system/process/process.go b/metricbeat/module/system/process/process.go
@@ -115,6 +115,8 @@ func (m *MetricSet) Fetch(r mb.ReporterV2) error {
 		if err != nil && !errors.Is(err, process.NonFatalErr{}) {
 			// return only if the error is fatal in nature
 			return fmt.Errorf("process stats: %w", err)
+		} else if (err != nil && errors.Is(err, process.NonFatalErr{})) {
+			err = mb.PartialMetricsError{Err: err}
 		}
 
 		for evtI := range procs {

diff --git a/metricbeat/module/system/process_summary/process_summary.go b/metricbeat/module/system/process_summary/process_summary.go
@@ -73,6 +73,8 @@ func (m *MetricSet) Fetch(r mb.ReporterV2) error {
 	if degradeErr != nil && !errors.Is(degradeErr, process.NonFatalErr{}) {
 		// return only if the error is fatal in nature
 		return fmt.Errorf("error fetching process list: %w", degradeErr)
+	} else if (degradeErr != nil && errors.Is(degradeErr, process.NonFatalErr{})) {
+		degradeErr = mb.PartialMetricsError{Err: degradeErr}
 	}
 
 	procStates := map[string]int{}