Merge pull request #194 from stefanotorresi/fix-cloned-resources-collection-once-and-for-all

fix cloned resource collection once and for all

Author: Stefano Torresi

Makefile

@@ -17,11 +17,7 @@ REPOSITORY ?= clusterlabs/ha_cluster_exporter
 # the Go archs we crosscompile to
 ARCHS ?= amd64 arm64 ppc64le s390x
 
-default: clean download mod-tidy generate fmt vet-check test build
-
-download:
-	go mod download
-	go mod verify
+default: clean mod-tidy generate fmt vet-check test build
 
 build: amd64
 
@@ -36,7 +32,7 @@ install:
 
 static-checks: vet-check fmt-check
 
-vet-check: download
+vet-check:
 	go vet ./...
 
 fmt:
@@ -51,7 +47,7 @@ fmt-check:
 generate:
 	go generate ./...
 
-test: download
+test:
 	go test -v ./...
 
 checks: static-checks test

collector/pacemaker/pacemaker.go

@@ -124,8 +124,16 @@ func (c *pacemakerCollector) recordResources(crmMon crmmon.Root, ch chan<- prome
 		c.recordResource(resource, "", "", ch)
 	}
 	for _, clone := range crmMon.Clones {
+		recorded := make(map[crmmon.Resource]bool) // we need to track cloned resources to avoid duplicates
 		for _, resource := range clone.Resources {
+			// Avoid recording stopped cloned resources multiple times
+			if recorded[resource] == true {
+				continue
+			}
+
 			c.recordResource(resource, "", clone.Id, ch)
+
+			recorded[resource] = true
 		}
 	}
 	for _, group := range crmMon.Groups {
@@ -146,9 +154,9 @@ func (c *pacemakerCollector) recordResource(resource crmmon.Resource, group stri
 		"failure_ignored": resource.FailureIgnored,
 	}
 
-	// Don't create metric for OCFS nodes that aren't running since those would generate duplicate entries
-	if resource.Agent == "ocf::heartbeat:Filesystem" && resource.NodesRunningOn == 0 {
-		return
+	var nodeName string
+	if resource.Node != nil {
+		nodeName = resource.Node.Name
 	}
 
 	// since we have a combined cardinality of resource * status, we cycle through all the possible statuses
@@ -158,21 +166,19 @@ func (c *pacemakerCollector) recordResource(resource crmmon.Resource, group stri
 		if flag {
 			statusValue = 1
 		}
-		var nodeName string
-		if resource.Node != nil {
-			nodeName = resource.Node.Name
-		}
-		ch <- c.MakeGaugeMetric(
-			"resources",
-			statusValue,
+
+		labels := []string{
 			nodeName,
 			resource.Id,
 			strings.ToLower(resource.Role),
 			strconv.FormatBool(resource.Managed),
 			resourceStatus,
 			resource.Agent,
 			group,
-			clone)
+			clone,
+		}
+
+		ch <- c.MakeGaugeMetric("resources", statusValue, labels...)
 	}
 }
 

test/mock_collector/instrumented_collector.go

@@ -103,6 +103,11 @@ ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-cluste
 ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="node02",resource="clusterfs",role="started",status="failed"} 0
 ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="node02",resource="clusterfs",role="started",status="failure_ignored"} 0
 ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="node02",resource="clusterfs",role="started",status="orphaned"} 0
+ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="",resource="clusterfs",role="stopped",status="active"} 0
+ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="",resource="clusterfs",role="stopped",status="blocked"} 0
+ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="",resource="clusterfs",role="stopped",status="failed"} 0
+ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="",resource="clusterfs",role="stopped",status="failure_ignored"} 0
+ha_cluster_pacemaker_resources{agent="ocf::heartbeat:Filesystem",clone="c-clusterfs",group="",managed="true",node="",resource="clusterfs",role="stopped",status="orphaned"} 0
 ha_cluster_pacemaker_resources{agent="ocf::heartbeat:IPaddr2",clone="",group="",managed="true",node="node01",resource="rsc_ip_PRD_HDB00",role="started",status="active"} 1
 ha_cluster_pacemaker_resources{agent="ocf::heartbeat:IPaddr2",clone="",group="",managed="true",node="node01",resource="rsc_ip_PRD_HDB00",role="started",status="blocked"} 0
 ha_cluster_pacemaker_resources{agent="ocf::heartbeat:IPaddr2",clone="",group="",managed="true",node="node01",resource="rsc_ip_PRD_HDB00",role="started",status="failed"} 0