telegraf/plugins/inputs/ceph/ceph.go

//go:generate ../../../tools/readme_config_includer/generator
package ceph

import (
	"bytes"
	_ "embed"
	"encoding/json"
	"fmt"
	"os"
	"os/exec"
	"path/filepath"
	"strings"

	"github.com/influxdata/telegraf"
	"github.com/influxdata/telegraf/plugins/inputs"
)

//go:embed sample.conf
var sampleConfig string

const (
	measurement = "ceph"
	typeMon     = "monitor"
	typeOsd     = "osd"
	typeMds     = "mds"
	typeRgw     = "rgw"
	osdPrefix   = "ceph-osd"
	monPrefix   = "ceph-mon"
	mdsPrefix   = "ceph-mds"
	rgwPrefix   = "ceph-client"
	sockSuffix  = "asok"
)

type Ceph struct {
	CephBinary             string `toml:"ceph_binary"`
	OsdPrefix              string `toml:"osd_prefix"`
	MonPrefix              string `toml:"mon_prefix"`
	MdsPrefix              string `toml:"mds_prefix"`
	RgwPrefix              string `toml:"rgw_prefix"`
	SocketDir              string `toml:"socket_dir"`
	SocketSuffix           string `toml:"socket_suffix"`
	CephUser               string `toml:"ceph_user"`
	CephConfig             string `toml:"ceph_config"`
	GatherAdminSocketStats bool   `toml:"gather_admin_socket_stats"`
	GatherClusterStats     bool   `toml:"gather_cluster_stats"`

	Log        telegraf.Logger `toml:"-"`
	schemaMaps map[socket]perfSchemaMap
}

func (*Ceph) SampleConfig() string {
	return sampleConfig
}

func (c *Ceph) Gather(acc telegraf.Accumulator) error {
	if c.GatherAdminSocketStats {
		if err := c.gatherAdminSocketStats(acc); err != nil {
			return err
		}
	}

	if c.GatherClusterStats {
		if err := c.gatherClusterStats(acc); err != nil {
			return err
		}
	}

	return nil
}

func (c *Ceph) gatherAdminSocketStats(acc telegraf.Accumulator) error {
	sockets, err := findSockets(c)
	if err != nil {
		return fmt.Errorf("failed to find sockets at path %q: %w", c.SocketDir, err)
	}

	if c.schemaMaps == nil {
		c.schemaMaps = make(map[socket]perfSchemaMap)
	}

	for _, s := range sockets {
		if _, ok := c.schemaMaps[*s]; !ok {
			rawSchema, err := perfSchema(c.CephBinary, s)
			if err != nil {
				c.Log.Warnf("failed to dump perf schema from socket %q: %v", s.socket, err)
			} else if schema, err := parseSchema(rawSchema); err != nil {
				c.Log.Warnf("failed to parse perf schema from socket %q: %v", s.socket, err)
			} else {
				c.schemaMaps[*s] = schema
			}
		}
		dump, err := perfDump(c.CephBinary, s)
		if err != nil {
			acc.AddError(fmt.Errorf("error reading from socket %q: %w", s.socket, err))
			continue
		}
		data, err := c.parseDump(dump)
		if err != nil {
			acc.AddError(fmt.Errorf("error parsing dump from socket %q: %w", s.socket, err))
			continue
		}
		for tag, metrics := range data {
			if schema, ok := c.schemaMaps[*s]; ok {
				for name, metric := range metrics {
					valueType := schema[tag][name]
					switch valueType {
					case telegraf.Counter:
						acc.AddCounter(measurement,
							map[string]interface{}{name: metric},
							map[string]string{"type": s.sockType, "id": s.sockID, "collection": tag})
					default:
						acc.AddGauge(measurement,
							map[string]interface{}{name: metric},
							map[string]string{"type": s.sockType, "id": s.sockID, "collection": tag})
					}
				}
			} else {
				acc.AddFields(measurement,
					metrics,
					map[string]string{"type": s.sockType, "id": s.sockID, "collection": tag})
			}
		}
	}
	return nil
}

func (c *Ceph) gatherClusterStats(acc telegraf.Accumulator) error {
	jobs := []struct {
		command string
		parser  func(telegraf.Accumulator, string) error
	}{
		{"status", decodeStatus},
		{"df", decodeDf},
		{"osd pool stats", decodeOsdPoolStats},
	}

	// For each job, execute against the cluster, parse and accumulate the data points
	for _, job := range jobs {
		output, err := c.execute(job.command)
		if err != nil {
			return fmt.Errorf("error executing command: %w", err)
		}
		err = job.parser(acc, output)
		if err != nil {
			return fmt.Errorf("error parsing output: %w", err)
		}
	}

	return nil
}

// Run ceph perf schema on the passed socket.  The output is a JSON string
// mapping collection names to a map of counter names to information.
//
// The counter information includes the type of the counter, which determines
// the names of the final series produced.  For example, a real-integer pair
// valued metric produces three series: sum, avgcount and avgtime; which hold
// the sum of all values, the count of all values and the division of these
// values.
func perfSchema(binary string, socket *socket) (string, error) {
	cmdArgs := []string{"--admin-daemon", socket.socket}

	switch socket.sockType {
	case typeOsd, typeMds, typeRgw:
		cmdArgs = append(cmdArgs, "perf", "schema")
	case typeMon:
		cmdArgs = append(cmdArgs, "perfcounters_schema")
	default:
		return "", fmt.Errorf("ignoring unknown socket type: %s", socket.sockType)
	}

	cmd := exec.Command(binary, cmdArgs...)
	var out bytes.Buffer
	cmd.Stdout = &out
	err := cmd.Run()
	if err != nil {
		return "", fmt.Errorf("error running ceph schema: %w", err)
	}

	return out.String(), nil
}

var perfDump = func(binary string, socket *socket) (string, error) {
	cmdArgs := []string{"--admin-daemon", socket.socket}

	switch socket.sockType {
	case typeOsd:
		cmdArgs = append(cmdArgs, "perf", "dump")
	case typeMon:
		cmdArgs = append(cmdArgs, "perfcounters_dump")
	case typeMds:
		cmdArgs = append(cmdArgs, "perf", "dump")
	case typeRgw:
		cmdArgs = append(cmdArgs, "perf", "dump")
	default:
		return "", fmt.Errorf("ignoring unknown socket type: %s", socket.sockType)
	}

	cmd := exec.Command(binary, cmdArgs...)
	var out bytes.Buffer
	cmd.Stdout = &out
	err := cmd.Run()
	if err != nil {
		return "", fmt.Errorf("error running ceph dump: %w", err)
	}

	return out.String(), nil
}

var findSockets = func(c *Ceph) ([]*socket, error) {
	listing, err := os.ReadDir(c.SocketDir)
	if err != nil {
		return nil, fmt.Errorf("failed to read socket directory %q: %w", c.SocketDir, err)
	}
	sockets := make([]*socket, 0, len(listing))
	for _, info := range listing {
		f := info.Name()
		var sockType string
		var sockPrefix string
		if strings.HasPrefix(f, c.MonPrefix) {
			sockType = typeMon
			sockPrefix = monPrefix
		}
		if strings.HasPrefix(f, c.OsdPrefix) {
			sockType = typeOsd
			sockPrefix = osdPrefix
		}
		if strings.HasPrefix(f, c.MdsPrefix) {
			sockType = typeMds
			sockPrefix = mdsPrefix
		}
		if strings.HasPrefix(f, c.RgwPrefix) {
			sockType = typeRgw
			sockPrefix = rgwPrefix
		}

		if sockType == typeOsd || sockType == typeMon || sockType == typeMds || sockType == typeRgw {
			path := filepath.Join(c.SocketDir, f)
			sockets = append(sockets, &socket{parseSockID(f, sockPrefix, c.SocketSuffix), sockType, path})
		}
	}
	return sockets, nil
}

func parseSockID(fname, prefix, suffix string) string {
	s := fname
	s = strings.TrimPrefix(s, prefix)
	s = strings.TrimSuffix(s, suffix)
	s = strings.Trim(s, ".-_")
	return s
}

type socket struct {
	sockID   string
	sockType string
	socket   string
}

type metric struct {
	pathStack []string // lifo stack of name components
	value     float64
}

// Pops names of pathStack to build the flattened name for a metric
func (m *metric) name() string {
	buf := bytes.Buffer{}
	for i := len(m.pathStack) - 1; i >= 0; i-- {
		if buf.Len() > 0 {
			buf.WriteString(".")
		}
		buf.WriteString(m.pathStack[i])
	}
	return buf.String()
}

type metricMap map[string]interface{}

type taggedMetricMap map[string]metricMap

// Mask bits for perf counters
const (
	perfCounterNone       = 0
	perfCounterTime       = 0x1
	perfCounterU64        = 0x2
	perfCounterLongRunAvg = 0x4
	perfCounterCounter    = 0x8
	perfCounterHistogram  = 0x10
)

type rawPerfCounter struct {
	TypeMask    int    `json:"type"`
	MetricType  string `json:"metric_type"`
	ValueType   string `json:"value_type"`
	Description string `json:"description"`
	Nick        string `json:"nick"`
	Priority    int    `json:"priority"`
	Units       string `json:"units"`
}

type rawCollection map[string]rawPerfCounter

type perfSchemaMap map[string]map[string]telegraf.ValueType

// Parses the output of ceph perf schema into a useful format, mapping metrics
// in collections to their Telegraf metric type.  This is made a little more
// complicated by the need to expand averages into their component metrics.
func parseSchema(rawSchema string) (perfSchemaMap, error) {
	rawMap := make(map[string]rawCollection)
	err := json.Unmarshal([]byte(rawSchema), &rawMap)
	if err != nil {
		return nil, fmt.Errorf("failed to parse json: %q: %w", rawSchema, err)
	}

	schemaMap := make(perfSchemaMap)
	for collection, counters := range rawMap {
		schemaMap[collection] = make(map[string]telegraf.ValueType)
		for counter, schema := range counters {
			if schema.TypeMask&perfCounterLongRunAvg != 0 {
				schemaMap[collection][counter+".sum"] = telegraf.Counter
				schemaMap[collection][counter+".avgcount"] = telegraf.Counter
				if schema.TypeMask&perfCounterTime != 0 {
					schemaMap[collection][counter+".avgtime"] = telegraf.Gauge
				}
			} else if schema.TypeMask&perfCounterCounter != 0 {
				schemaMap[collection][counter] = telegraf.Counter
			} else {
				schemaMap[collection][counter] = telegraf.Gauge
			}
		}
	}
	return schemaMap, nil
}

// Parses a raw JSON string into a taggedMetricMap
// Delegates the actual parsing to newTaggedMetricMap(..)
func (c *Ceph) parseDump(dump string) (taggedMetricMap, error) {
	data := make(map[string]interface{})
	err := json.Unmarshal([]byte(dump), &data)
	if err != nil {
		return nil, fmt.Errorf("failed to parse json: %q: %w", dump, err)
	}

	return c.newTaggedMetricMap(data), nil
}

// Builds a TaggedMetricMap out of a generic string map.
// The top-level key is used as a tag and all sub-keys are flattened into metrics
func (c *Ceph) newTaggedMetricMap(data map[string]interface{}) taggedMetricMap {
	tmm := make(taggedMetricMap)
	for tag, datapoints := range data {
		mm := make(metricMap)
		for _, m := range c.flatten(datapoints) {
			mm[m.name()] = m.value
		}
		tmm[tag] = mm
	}
	return tmm
}

// Recursively flattens any k-v hierarchy present in data.
// Nested keys are flattened into ordered slices associated with a metric value.
// The key slices are treated as stacks, and are expected to be reversed and concatenated
// when passed as metrics to the accumulator. (see (*metric).name())
func (c *Ceph) flatten(data interface{}) []*metric {
	var metrics []*metric

	switch val := data.(type) {
	case float64:
		metrics = []*metric{
			{
				make([]string, 0, 1), val,
			},
		}
	case map[string]interface{}:
		metrics = make([]*metric, 0, len(val))
		for k, v := range val {
			for _, m := range c.flatten(v) {
				m.pathStack = append(m.pathStack, k)
				metrics = append(metrics, m)
			}
		}
	default:
		c.Log.Infof("ignoring unexpected type '%T' for value %v", val, val)
	}

	return metrics
}

// execute executes the 'ceph' command with the supplied arguments, returning JSON formatted output
func (c *Ceph) execute(command string) (string, error) {
	cmdArgs := []string{"--conf", c.CephConfig, "--name", c.CephUser, "--format", "json"}
	cmdArgs = append(cmdArgs, strings.Split(command, " ")...)

	cmd := exec.Command(c.CephBinary, cmdArgs...)

	var out bytes.Buffer
	cmd.Stdout = &out
	err := cmd.Run()
	if err != nil {
		return "", fmt.Errorf("error running ceph %q: %w", command, err)
	}

	output := out.String()

	// Ceph doesn't sanitize its output, and may return invalid JSON.  Patch this
	// up for them, as having some inaccurate data is better than none.
	output = strings.ReplaceAll(output, "-inf", "0")
	output = strings.ReplaceAll(output, "inf", "0")

	return output, nil
}

// status is used to unmarshal "ceph -s" output
type status struct {
	FSMap struct {
		NumIn        float64 `json:"in"`
		NumMax       float64 `json:"max"`
		NumUp        float64 `json:"up"`
		NumUpStandby float64 `json:"up:standby"`
	} `json:"fsmap"`
	Health struct {
		OverallStatus string `json:"overall_status"` // Only valid for ceph version <15
		Status        string `json:"status"`
	} `json:"health"`
	MonMap struct {
		NumMons float64 `json:"num_mons"`
	} `json:"monmap"`
	OSDMap struct {
		Epoch          float64  `json:"epoch"`
		NumInOSDs      float64  `json:"num_in_osds"`
		NumOSDs        float64  `json:"num_osds"`
		NumRemappedPGs float64  `json:"num_remapped_pgs"`
		NumUpOSDs      float64  `json:"num_up_osds"`
		OSDMap         struct { // nested OSDmap used in ceph version <15
			Epoch          float64 `json:"epoch"`
			Full           bool    `json:"full"`
			NearFull       bool    `json:"nearfull"`
			NumInOSDs      float64 `json:"num_in_osds"`
			NumOSDs        float64 `json:"num_osds"`
			NumRemappedPGs float64 `json:"num_remapped_pgs"`
			NumUpOSDs      float64 `json:"num_up_osds"`
		} `json:"osdmap"`
	} `json:"osdmap"`
	PGMap struct {
		PGsByState []struct {
			Count     float64 `json:"count"`
			StateName string  `json:"state_name"`
		} `json:"pgs_by_state"`
		BytesAvail              float64 `json:"bytes_avail"`
		BytesTotal              float64 `json:"bytes_total"`
		BytesUsed               float64 `json:"bytes_used"`
		DataBytes               float64 `json:"data_bytes"`
		DegradedObjects         float64 `json:"degraded_objects"`
		DegradedRatio           float64 `json:"degraded_ratio"`
		DegradedTotal           float64 `json:"degraded_total"`
		InactivePGsRatio        float64 `json:"inactive_pgs_ratio"`
		NumBytesRecovered       float64 `json:"num_bytes_recovered"`
		NumKeysRecovered        float64 `json:"num_keys_recovered"`
		NumObjects              float64 `json:"num_objects"`
		NumObjectRecovered      float64 `json:"num_objects_recovered"`
		NumPGs                  float64 `json:"num_pgs"`
		NumPools                float64 `json:"num_pools"`
		OpPerSec                float64 `json:"op_per_sec"` // This field is no longer reported in ceph 10 and later
		ReadBytesSec            float64 `json:"read_bytes_sec"`
		ReadOpPerSec            float64 `json:"read_op_per_sec"`
		RecoveringBytesPerSec   float64 `json:"recovering_bytes_per_sec"`
		RecoveringKeysPerSec    float64 `json:"recovering_keys_per_sec"`
		RecoveringObjectsPerSec float64 `json:"recovering_objects_per_sec"`
		Version                 float64 `json:"version"`
		WriteBytesSec           float64 `json:"write_bytes_sec"`
		WriteOpPerSec           float64 `json:"write_op_per_sec"`
	} `json:"pgmap"`
}

// decodeStatus decodes the output of 'ceph -s'
func decodeStatus(acc telegraf.Accumulator, input string) error {
	data := &status{}
	if err := json.Unmarshal([]byte(input), data); err != nil {
		return fmt.Errorf("failed to parse json: %q: %w", input, err)
	}

	decoders := []func(telegraf.Accumulator, *status) error{
		decodeStatusFsmap,
		decodeStatusHealth,
		decodeStatusMonmap,
		decodeStatusOsdmap,
		decodeStatusPgmap,
		decodeStatusPgmapState,
	}

	for _, decoder := range decoders {
		if err := decoder(acc, data); err != nil {
			return err
		}
	}

	return nil
}

// decodeStatusFsmap decodes the FS map portion of the output of 'ceph -s'
func decodeStatusFsmap(acc telegraf.Accumulator, data *status) error {
	fields := map[string]interface{}{
		"in":         data.FSMap.NumIn,
		"max":        data.FSMap.NumMax,
		"up_standby": data.FSMap.NumUpStandby,
		"up":         data.FSMap.NumUp,
	}
	acc.AddFields("ceph_fsmap", fields, make(map[string]string))
	return nil
}

// decodeStatusHealth decodes the health portion of the output of 'ceph status'
func decodeStatusHealth(acc telegraf.Accumulator, data *status) error {
	statusCodes := map[string]float64{
		"HEALTH_ERR":  0,
		"HEALTH_WARN": 1,
		"HEALTH_OK":   2,
	}
	fields := map[string]interface{}{
		"overall_status": data.Health.OverallStatus, // This field is no longer reported in ceph 10 and later
		"status_code":    statusCodes[data.Health.Status],
		"status":         data.Health.Status,
	}
	acc.AddFields("ceph_health", fields, make(map[string]string))
	return nil
}

// decodeStatusMonmap decodes the Mon map portion of the output of 'ceph -s'
func decodeStatusMonmap(acc telegraf.Accumulator, data *status) error {
	fields := map[string]interface{}{
		"num_mons": data.MonMap.NumMons,
	}
	acc.AddFields("ceph_monmap", fields, make(map[string]string))
	return nil
}

// decodeStatusOsdmap decodes the OSD map portion of the output of 'ceph -s'
func decodeStatusOsdmap(acc telegraf.Accumulator, data *status) error {
	fields := map[string]interface{}{
		"epoch":            data.OSDMap.Epoch,
		"num_in_osds":      data.OSDMap.NumInOSDs,
		"num_osds":         data.OSDMap.NumOSDs,
		"num_remapped_pgs": data.OSDMap.NumRemappedPGs,
		"num_up_osds":      data.OSDMap.NumUpOSDs,
	}
	if data.OSDMap.OSDMap.Epoch != 0 && data.OSDMap.OSDMap.NumOSDs != 0 {
		fields = map[string]interface{}{
			"epoch":            data.OSDMap.OSDMap.Epoch,
			"full":             data.OSDMap.OSDMap.Full,
			"nearfull":         data.OSDMap.OSDMap.NearFull,
			"num_in_osds":      data.OSDMap.OSDMap.NumInOSDs,
			"num_osds":         data.OSDMap.OSDMap.NumOSDs,
			"num_remapped_pgs": data.OSDMap.OSDMap.NumRemappedPGs,
			"num_up_osds":      data.OSDMap.OSDMap.NumUpOSDs,
		}
	}

	acc.AddFields("ceph_osdmap", fields, make(map[string]string))
	return nil
}

// decodeStatusPgmap decodes the PG map portion of the output of 'ceph -s'
func decodeStatusPgmap(acc telegraf.Accumulator, data *status) error {
	fields := map[string]interface{}{
		"bytes_avail":                data.PGMap.BytesAvail,
		"bytes_total":                data.PGMap.BytesTotal,
		"bytes_used":                 data.PGMap.BytesUsed,
		"data_bytes":                 data.PGMap.DataBytes,
		"degraded_objects":           data.PGMap.DegradedObjects,
		"degraded_ratio":             data.PGMap.DegradedRatio,
		"degraded_total":             data.PGMap.DegradedTotal,
		"inactive_pgs_ratio":         data.PGMap.InactivePGsRatio,
		"num_bytes_recovered":        data.PGMap.NumBytesRecovered,
		"num_keys_recovered":         data.PGMap.NumKeysRecovered,
		"num_objects_recovered":      data.PGMap.NumObjectRecovered,
		"num_objects":                data.PGMap.NumObjects,
		"num_pgs":                    data.PGMap.NumPGs,
		"num_pools":                  data.PGMap.NumPools,
		"op_per_sec":                 data.PGMap.OpPerSec, // This field is no longer reported in ceph 10 and later
		"read_bytes_sec":             data.PGMap.ReadBytesSec,
		"read_op_per_sec":            data.PGMap.ReadOpPerSec,
		"recovering_bytes_per_sec":   data.PGMap.RecoveringBytesPerSec,
		"recovering_keys_per_sec":    data.PGMap.RecoveringKeysPerSec,
		"recovering_objects_per_sec": data.PGMap.RecoveringObjectsPerSec,
		"version":                    data.PGMap.Version,
		"write_bytes_sec":            data.PGMap.WriteBytesSec,
		"write_op_per_sec":           data.PGMap.WriteOpPerSec,
	}
	acc.AddFields("ceph_pgmap", fields, make(map[string]string))
	return nil
}

// decodeStatusPgmapState decodes the PG map state portion of the output of 'ceph -s'
func decodeStatusPgmapState(acc telegraf.Accumulator, data *status) error {
	for _, pgState := range data.PGMap.PGsByState {
		tags := map[string]string{
			"state": pgState.StateName,
		}
		fields := map[string]interface{}{
			"count": pgState.Count,
		}
		acc.AddFields("ceph_pgmap_state", fields, tags)
	}
	return nil
}

// df is used to unmarshal 'ceph df' output
type df struct {
	Stats struct {
		NumOSDs            float64 `json:"num_osds"`
		NumPerPoolOmapOSDs float64 `json:"num_per_pool_omap_osds"`
		NumPerPoolOSDs     float64 `json:"num_per_pool_osds"`
		TotalAvail         float64 `json:"total_avail"` // pre ceph 0.84
		TotalAvailBytes    float64 `json:"total_avail_bytes"`
		TotalBytes         float64 `json:"total_bytes"`
		TotalSpace         float64 `json:"total_space"` // pre ceph 0.84
		TotalUsed          float64 `json:"total_used"`  // pre ceph 0.84
		TotalUsedBytes     float64 `json:"total_used_bytes"`
		TotalUsedRawBytes  float64 `json:"total_used_raw_bytes"`
		TotalUsedRawRatio  float64 `json:"total_used_raw_ratio"`
	} `json:"stats"`
	StatsbyClass map[string]map[string]float64 `json:"stats_by_class"`
	Pools        []struct {
		Name  string `json:"name"`
		Stats struct {
			BytesUsed   float64 `json:"bytes_used"`
			KBUsed      float64 `json:"kb_used"`
			MaxAvail    float64 `json:"max_avail"`
			Objects     float64 `json:"objects"`
			PercentUsed float64 `json:"percent_used"`
			Stored      float64 `json:"stored"`
		} `json:"stats"`
	} `json:"pools"`
}

// decodeDf decodes the output of 'ceph df'
func decodeDf(acc telegraf.Accumulator, input string) error {
	data := &df{}
	if err := json.Unmarshal([]byte(input), data); err != nil {
		return fmt.Errorf("failed to parse json: %q: %w", input, err)
	}

	// ceph.usage: records global utilization and number of objects
	fields := map[string]interface{}{
		"num_osds":               data.Stats.NumOSDs,
		"num_per_pool_omap_osds": data.Stats.NumPerPoolOmapOSDs,
		"num_per_pool_osds":      data.Stats.NumPerPoolOSDs,
		"total_avail_bytes":      data.Stats.TotalAvailBytes,
		"total_avail":            data.Stats.TotalAvail, // pre ceph 0.84
		"total_bytes":            data.Stats.TotalBytes,
		"total_space":            data.Stats.TotalSpace, // pre ceph 0.84
		"total_used_bytes":       data.Stats.TotalUsedBytes,
		"total_used_raw_bytes":   data.Stats.TotalUsedRawBytes,
		"total_used_raw_ratio":   data.Stats.TotalUsedRawRatio,
		"total_used":             data.Stats.TotalUsed, // pre ceph 0.84
	}
	acc.AddFields("ceph_usage", fields, make(map[string]string))

	// ceph.stats_by_class: records per device-class usage
	for class, stats := range data.StatsbyClass {
		tags := map[string]string{
			"class": class,
		}
		fields := make(map[string]interface{})
		for key, value := range stats {
			fields[key] = value
		}
		acc.AddFields("ceph_deviceclass_usage", fields, tags)
	}

	// ceph.pool.usage: records per pool utilization and number of objects
	for _, pool := range data.Pools {
		tags := map[string]string{
			"name": pool.Name,
		}
		fields := map[string]interface{}{
			"bytes_used":   pool.Stats.BytesUsed,
			"kb_used":      pool.Stats.KBUsed,
			"max_avail":    pool.Stats.MaxAvail,
			"objects":      pool.Stats.Objects,
			"percent_used": pool.Stats.PercentUsed,
			"stored":       pool.Stats.Stored,
		}
		acc.AddFields("ceph_pool_usage", fields, tags)
	}

	return nil
}

// osdPoolStats is used to unmarshal 'ceph osd pool stats' output
type osdPoolStats []struct {
	PoolName     string `json:"pool_name"`
	ClientIORate struct {
		OpPerSec      float64 `json:"op_per_sec"` // This field is no longer reported in ceph 10 and later
		ReadBytesSec  float64 `json:"read_bytes_sec"`
		ReadOpPerSec  float64 `json:"read_op_per_sec"`
		WriteBytesSec float64 `json:"write_bytes_sec"`
		WriteOpPerSec float64 `json:"write_op_per_sec"`
	} `json:"client_io_rate"`
	RecoveryRate struct {
		NumBytesRecovered       float64 `json:"num_bytes_recovered"`
		NumKeysRecovered        float64 `json:"num_keys_recovered"`
		NumObjectRecovered      float64 `json:"num_objects_recovered"`
		RecoveringBytesPerSec   float64 `json:"recovering_bytes_per_sec"`
		RecoveringKeysPerSec    float64 `json:"recovering_keys_per_sec"`
		RecoveringObjectsPerSec float64 `json:"recovering_objects_per_sec"`
	} `json:"recovery_rate"`
	Recovery struct {
		DegradedObjects float64 `json:"degraded_objects"`
		DegradedRatio   float64 `json:"degraded_ratio"`
		DegradedTotal   float64 `json:"degraded_total"`
	} `json:"recovery"`
}

// decodeOsdPoolStats decodes the output of 'ceph osd pool stats'
func decodeOsdPoolStats(acc telegraf.Accumulator, input string) error {
	data := make(osdPoolStats, 0)
	if err := json.Unmarshal([]byte(input), &data); err != nil {
		return fmt.Errorf("failed to parse json: %q: %w", input, err)
	}

	// ceph.pool.stats: records pre pool IO and recovery throughput
	for _, pool := range data {
		tags := map[string]string{
			"name": pool.PoolName,
		}
		fields := map[string]interface{}{
			"degraded_objects":           pool.Recovery.DegradedObjects,
			"degraded_ratio":             pool.Recovery.DegradedRatio,
			"degraded_total":             pool.Recovery.DegradedTotal,
			"num_bytes_recovered":        pool.RecoveryRate.NumBytesRecovered,
			"num_keys_recovered":         pool.RecoveryRate.NumKeysRecovered,
			"num_objects_recovered":      pool.RecoveryRate.NumObjectRecovered,
			"op_per_sec":                 pool.ClientIORate.OpPerSec, // This field is no longer reported in ceph 10 and later
			"read_bytes_sec":             pool.ClientIORate.ReadBytesSec,
			"read_op_per_sec":            pool.ClientIORate.ReadOpPerSec,
			"recovering_bytes_per_sec":   pool.RecoveryRate.RecoveringBytesPerSec,
			"recovering_keys_per_sec":    pool.RecoveryRate.RecoveringKeysPerSec,
			"recovering_objects_per_sec": pool.RecoveryRate.RecoveringObjectsPerSec,
			"write_bytes_sec":            pool.ClientIORate.WriteBytesSec,
			"write_op_per_sec":           pool.ClientIORate.WriteOpPerSec,
		}
		acc.AddFields("ceph_pool_stats", fields, tags)
	}

	return nil
}

func init() {
	inputs.Add(measurement, func() telegraf.Input {
		return &Ceph{
			CephBinary:             "/usr/bin/ceph",
			OsdPrefix:              osdPrefix,
			MonPrefix:              monPrefix,
			MdsPrefix:              mdsPrefix,
			RgwPrefix:              rgwPrefix,
			SocketDir:              "/var/run/ceph",
			SocketSuffix:           sockSuffix,
			CephUser:               "client.admin",
			CephConfig:             "/etc/ceph/ceph.conf",
			GatherAdminSocketStats: true,
			GatherClusterStats:     false,
		}
	})
}