From 47b7ae2ed638587cd87b6de7397359d2406a635d Mon Sep 17 00:00:00 2001
From: Matt Vandersomething <matt1360@users.noreply.github.com>
Date: Tue, 21 Dec 2021 14:24:47 -0400
Subject: [PATCH] collectors/health: add repair state checking

---
 collectors/health.go      | 15 +++++++++++++++
 collectors/health_test.go |  5 +++++
 2 files changed, 20 insertions(+)

diff --git a/collectors/health.go b/collectors/health.go
index 3a9b0f8..701ab65 100644
--- a/collectors/health.go
+++ b/collectors/health.go
@@ -158,6 +158,9 @@ type ClusterHealthCollector struct {
 	// SnaptrimWaitPGs depicts no. of PGs that are currently waiting to snaptrim
 	SnaptrimWaitPGs prometheus.Gauge
 
+	// RepairingPGs depicts no. of PGs that are currently repairing
+	RepairingPGs prometheus.Gauge
+
 	// SlowOps depicts no. of total slow ops in the cluster
 	SlowOps prometheus.Gauge
 
@@ -507,6 +510,14 @@ func NewClusterHealthCollector(conn Conn, cluster string, logger *logrus.Logger)
 				ConstLabels: labels,
 			},
 		),
+		RepairingPGs: prometheus.NewGauge(
+			prometheus.GaugeOpts{
+				Namespace:   cephNamespace,
+				Name:        "repairing_pgs",
+				Help:        "No. of PGs in the cluster with repair state",
+				ConstLabels: labels,
+			},
+		),
 		// with Nautilus, SLOW_OPS has replaced both REQUEST_SLOW and REQUEST_STUCK
 		// therefore slow_requests is deprecated, but for backwards compatibility
 		// the metric name will be kept the same for the time being
@@ -916,6 +927,7 @@ func (c *ClusterHealthCollector) metricsList() []prometheus.Metric {
 		c.InconsistentPGs,
 		c.SnaptrimPGs,
 		c.SnaptrimWaitPGs,
+		c.RepairingPGs,
 		c.SlowOps,
 		c.DegradedObjectsCount,
 		c.MisplacedObjectsCount,
@@ -1252,6 +1264,7 @@ func (c *ClusterHealthCollector) collect(ch chan<- prometheus.Metric) error {
 		inconsistentPGs   float64
 		snaptrimPGs       float64
 		snaptrimWaitPGs   float64
+		repairingPGs      float64
 
 		pgStateCounterMap = map[string]*float64{
 			"degraded":        &degradedPGs,
@@ -1273,6 +1286,7 @@ func (c *ClusterHealthCollector) collect(ch chan<- prometheus.Metric) error {
 			"inconsistent":    &inconsistentPGs,
 			"snaptrim":        &snaptrimPGs,
 			"snaptrim_wait":   &snaptrimWaitPGs,
+			"repair":          &repairingPGs,
 		}
 		pgStateGaugeMap = map[string]prometheus.Gauge{
 			"degraded":        c.DegradedPGs,
@@ -1294,6 +1308,7 @@ func (c *ClusterHealthCollector) collect(ch chan<- prometheus.Metric) error {
 			"inconsistent":    c.InconsistentPGs,
 			"snaptrim":        c.SnaptrimPGs,
 			"snaptrim_wait":   c.SnaptrimWaitPGs,
+			"repair":          c.RepairingPGs,
 		}
 	)
 
diff --git a/collectors/health_test.go b/collectors/health_test.go
index 64aba9c..876f6cf 100644
--- a/collectors/health_test.go
+++ b/collectors/health_test.go
@@ -340,6 +340,10 @@ $ sudo ceph -s
 				"state_name": "active+clean+inconsistent",
 				"count": 1
 			},
+			{
+				"state_name": "active+clean+repair",
+				"count": 1
+			},
 			{
 				"state_name": "active+clean+snaptrim",
 				"count": 15
@@ -362,6 +366,7 @@ $ sudo ceph -s
 				regexp.MustCompile(`cluster_objects{cluster="ceph"} 13156`),
 				regexp.MustCompile(`snaptrim_pgs{cluster="ceph"} 15`),
 				regexp.MustCompile(`snaptrim_wait_pgs{cluster="ceph"} 25`),
+				regexp.MustCompile(`repair{cluster="ceph"} 1`),
 			},
 		},
 		{