PMM-12384 Backup artifact metrics (#2428)

* PMM-12384 Add backup artifact metrics

* PMM-12384 Fixes

* PMM_12384 Fix

* PMM-12384 Remove TODO

* PMM-12384 Refactoring

* Update managed/data/iatemplates/backup_error.yml

Author: artemgavrilov

dev/mongo-rs-backups/docker-compose.yml

@@ -1,6 +1,7 @@
 networks:
   pmm_default:
     name: pmm_default
+    external: true
 
 services:
   mongo1:

managed/cmd/pmm-managed/main.go

@@ -932,6 +932,9 @@ func main() { //nolint:cyclop,maintidx
 	dbaasClient := dbaas.NewClient(*dbaasControllerAPIAddrF)
 	compatibilityService := backup.NewCompatibilityService(db, versioner)
 	backupService := backup.NewService(db, jobsService, agentService, compatibilityService, pbmPITRService)
+	backupMetricsCollector := backup.NewMetricsCollector(db)
+	prom.MustRegister(backupMetricsCollector)
+
 	schedulerService := scheduler.New(db, backupService)
 	versionCache := versioncache.New(db, versioner)
 	emailer := alertmanager.NewEmailer(logrus.WithField("component", "alertmanager-emailer").Logger)

managed/data/iatemplates/backup_error.yml

@@ -0,0 +1,12 @@
+---
+templates:
+  - name: pmm_backup_error
+    version: 1
+    summary: Backup failed (Tech preview)
+    expr: 'pmm_managed_backups_artifacts{status="error"} == bool 1'
+    for: 1m
+    severity: error
+    annotations:
+      description: |-
+        Failed to create a backup artifact '{{ $labels.artifact_name}}' on service '{{ $labels.service_id }}'.
+      summary: Failed to create a backup artifact '{{ $labels.artifact_name}}' on service '{{ $labels.service_id }}'.

managed/services/backup/metrics.go

@@ -0,0 +1,113 @@
+// Copyright (C) 2017 Percona LLC
+//
+// This program is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Affero General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+// GNU Affero General Public License for more details.
+//
+// You should have received a copy of the GNU Affero General Public License
+// along with this program. If not, see <https://www.gnu.org/licenses/>.
+
+package backup
+
+import (
+	"context"
+	"time"
+
+	"github.com/pkg/errors"
+	prom "github.com/prometheus/client_golang/prometheus"
+	"github.com/sirupsen/logrus"
+	"gopkg.in/reform.v1"
+
+	"github.com/percona/pmm/managed/models"
+)
+
+const (
+	requestTimeout              = 3 * time.Second
+	artifactExists      float64 = 1
+	prometheusNamespace         = "pmm_managed"
+	prometheusSubsystem         = "backups"
+)
+
+type MetricsCollector struct {
+	db *reform.DB
+	l  *logrus.Entry
+
+	mArtifactsDesc *prom.Desc
+}
+
+func NewMetricsCollector(db *reform.DB) *MetricsCollector {
+	return &MetricsCollector{
+		db: db,
+		l:  logrus.WithField("component", "backups/metrics"),
+		mArtifactsDesc: prom.NewDesc(
+			prom.BuildFQName(prometheusNamespace, prometheusSubsystem, "artifacts"),
+			"Artifacts",
+			[]string{
+				"artifact_id", "artifact_name", "artifact_vendor", "service_id", "service_name",
+				"type", "db_version", "data_model", "mode", "status",
+			},
+			nil),
+	}
+}
+
+func (c *MetricsCollector) Describe(ch chan<- *prom.Desc) {
+	ch <- c.mArtifactsDesc
+}
+
+func (c *MetricsCollector) Collect(ch chan<- prom.Metric) {
+	ctx, cancelCtx := context.WithTimeout(context.Background(), requestTimeout)
+	defer cancelCtx()
+
+	var artifacts []*models.Artifact
+	var services map[string]*models.Service
+	errTx := c.db.InTransactionContext(ctx, nil, func(t *reform.TX) error {
+		var err error
+		artifacts, err = models.FindArtifacts(t.Querier, models.ArtifactFilters{})
+		if err != nil {
+			return errors.WithStack(err)
+		}
+
+		serviceIDs := make([]string, 0, len(artifacts))
+		for _, artifact := range artifacts {
+			serviceIDs = append(serviceIDs, artifact.ServiceID)
+		}
+
+		services, err = models.FindServicesByIDs(t.Querier, serviceIDs)
+		if err != nil {
+			return errors.WithStack(err)
+		}
+		return nil
+	})
+	if errTx != nil {
+		c.l.Warnf("Failed to get artifacts: %v", errTx)
+		return
+	}
+
+	for _, artifact := range artifacts {
+		var serviceName string
+		if service, ok := services[artifact.ServiceID]; ok {
+			serviceName = service.ServiceName
+		}
+
+		ch <- prom.MustNewConstMetric(
+			c.mArtifactsDesc,
+			prom.GaugeValue,
+			artifactExists,
+			artifact.ID,
+			artifact.Name,
+			artifact.Vendor,
+			artifact.ServiceID,
+			serviceName,
+			string(artifact.Type),
+			artifact.DBVersion,
+			string(artifact.DataModel),
+			string(artifact.Mode),
+			string(artifact.Status))
+	}
+}