Add a collector for ZFS, currently focussed on ARC stats.

It is tested on FreeBSD 10.2-RELEASE and Linux (ZFS on Linux 0.6.5.4).

On FreeBSD, Solaris, etc. ZFS metrics are exposed through sysctls.
ZFS on Linux exposes the same metrics through procfs `/proc/spl/...`.

In addition to sysctl metrics, 'computed metrics' are exposed by
the collector, which are based on several sysctl values.
There is some conditional logic involved in computing these metrics
which cannot be easily mapped to PromQL.

Not all 92 ARC sysctls are exposed right now but this can be changed
with one additional LOC each.

collector/fixtures/proc/spl/kstat/zfs/arcstats

@@ -0,0 +1,93 @@
+6 1 0x01 91 4368 5266997922 97951858082072
+name                            type data
+hits                            4    8772612
+misses                          4    604635
+demand_data_hits                4    7221032
+demand_data_misses              4    73300
+demand_metadata_hits            4    1464353
+demand_metadata_misses          4    498170
+prefetch_data_hits              4    3615
+prefetch_data_misses            4    17094
+prefetch_metadata_hits          4    83612
+prefetch_metadata_misses        4    16071
+mru_hits                        4    855535
+mru_ghost_hits                  4    21100
+mfu_hits                        4    7829854
+mfu_ghost_hits                  4    821
+deleted                         4    60403
+mutex_miss                      4    2
+evict_skip                      4    2265729
+evict_not_enough                4    680
+evict_l2_cached                 4    0
+evict_l2_eligible               4    8992514560
+evict_l2_ineligible             4    992552448
+evict_l2_skip                   4    0
+hash_elements                   4    42359
+hash_elements_max               4    88245
+hash_collisions                 4    50564
+hash_chains                     4    412
+hash_chain_max                  4    3
+p                               4    516395305
+c                               4    1643208777
+c_min                           4    33554432
+c_max                           4    8367976448
+size                            4    1603939792
+hdr_size                        4    16361080
+data_size                       4    1295836160
+metadata_size                   4    175298560
+other_size                      4    116443992
+anon_size                       4    1917440
+anon_evictable_data             4    0
+anon_evictable_metadata         4    0
+mru_size                        4    402593792
+mru_evictable_data              4    278091264
+mru_evictable_metadata          4    18606592
+mru_ghost_size                  4    999728128
+mru_ghost_evictable_data        4    883765248
+mru_ghost_evictable_metadata    4    115962880
+mfu_size                        4    1066623488
+mfu_evictable_data              4    1017613824
+mfu_evictable_metadata          4    9163776
+mfu_ghost_size                  4    104936448
+mfu_ghost_evictable_data        4    96731136
+mfu_ghost_evictable_metadata    4    8205312
+l2_hits                         4    0
+l2_misses                       4    0
+l2_feeds                        4    0
+l2_rw_clash                     4    0
+l2_read_bytes                   4    0
+l2_write_bytes                  4    0
+l2_writes_sent                  4    0
+l2_writes_done                  4    0
+l2_writes_error                 4    0
+l2_writes_lock_retry            4    0
+l2_evict_lock_retry             4    0
+l2_evict_reading                4    0
+l2_evict_l1cached               4    0
+l2_free_on_write                4    0
+l2_cdata_free_on_write          4    0
+l2_abort_lowmem                 4    0
+l2_cksum_bad                    4    0
+l2_io_error                     4    0
+l2_size                         4    0
+l2_asize                        4    0
+l2_hdr_size                     4    0
+l2_compress_successes           4    0
+l2_compress_zeros               4    0
+l2_compress_failures            4    0
+memory_throttle_count           4    0
+duplicate_buffers               4    0
+duplicate_buffers_size          4    0
+duplicate_reads                 4    0
+memory_direct_count             4    542
+memory_indirect_count           4    3006
+arc_no_grow                     4    0
+arc_tempreserve                 4    0
+arc_loaned_bytes                4    0
+arc_prune                       4    0
+arc_meta_used                   4    308103632
+arc_meta_limit                  4    6275982336
+arc_meta_max                    4    449286096
+arc_meta_min                    4    16777216
+arc_need_free                   4    0
+arc_sys_free                    4    261496832

collector/fixtures/sysctl/freebsd/kstat.zfs.misc.arcstats.txt

@@ -0,0 +1,78 @@
+kstat.zfs.misc.arcstats.arc_meta_max: 1503210048
+kstat.zfs.misc.arcstats.arc_meta_limit: 393216000
+kstat.zfs.misc.arcstats.arc_meta_used: 392649848
+kstat.zfs.misc.arcstats.duplicate_reads: 0
+kstat.zfs.misc.arcstats.duplicate_buffers_size: 0
+kstat.zfs.misc.arcstats.duplicate_buffers: 0
+kstat.zfs.misc.arcstats.memory_throttle_count: 0
+kstat.zfs.misc.arcstats.l2_write_buffer_list_null_iter: 0
+kstat.zfs.misc.arcstats.l2_write_buffer_list_iter: 0
+kstat.zfs.misc.arcstats.l2_write_buffer_bytes_scanned: 0
+kstat.zfs.misc.arcstats.l2_write_pios: 0
+kstat.zfs.misc.arcstats.l2_write_buffer_iter: 0
+kstat.zfs.misc.arcstats.l2_write_full: 0
+kstat.zfs.misc.arcstats.l2_write_not_cacheable: 29425
+kstat.zfs.misc.arcstats.l2_write_io_in_progress: 0
+kstat.zfs.misc.arcstats.l2_write_in_l2: 0
+kstat.zfs.misc.arcstats.l2_write_spa_mismatch: 0
+kstat.zfs.misc.arcstats.l2_write_passed_headroom: 0
+kstat.zfs.misc.arcstats.l2_write_trylock_fail: 0
+kstat.zfs.misc.arcstats.l2_compress_failures: 0
+kstat.zfs.misc.arcstats.l2_compress_zeros: 0
+kstat.zfs.misc.arcstats.l2_compress_successes: 0
+kstat.zfs.misc.arcstats.l2_hdr_size: 0
+kstat.zfs.misc.arcstats.l2_asize: 0
+kstat.zfs.misc.arcstats.l2_size: 0
+kstat.zfs.misc.arcstats.l2_io_error: 0
+kstat.zfs.misc.arcstats.l2_cksum_bad: 0
+kstat.zfs.misc.arcstats.l2_abort_lowmem: 0
+kstat.zfs.misc.arcstats.l2_cdata_free_on_write: 0
+kstat.zfs.misc.arcstats.l2_free_on_write: 0
+kstat.zfs.misc.arcstats.l2_evict_reading: 0
+kstat.zfs.misc.arcstats.l2_evict_lock_retry: 0
+kstat.zfs.misc.arcstats.l2_writes_hdr_miss: 0
+kstat.zfs.misc.arcstats.l2_writes_error: 0
+kstat.zfs.misc.arcstats.l2_writes_done: 0
+kstat.zfs.misc.arcstats.l2_writes_sent: 0
+kstat.zfs.misc.arcstats.l2_write_bytes: 0
+kstat.zfs.misc.arcstats.l2_read_bytes: 0
+kstat.zfs.misc.arcstats.l2_rw_clash: 0
+kstat.zfs.misc.arcstats.l2_feeds: 0
+kstat.zfs.misc.arcstats.l2_misses: 0
+kstat.zfs.misc.arcstats.l2_hits: 0
+kstat.zfs.misc.arcstats.other_size: 166832272
+kstat.zfs.misc.arcstats.data_size: 1200779776
+kstat.zfs.misc.arcstats.hdr_size: 27244008
+kstat.zfs.misc.arcstats.size: 1394856056
+kstat.zfs.misc.arcstats.c_max: 1572864000
+kstat.zfs.misc.arcstats.c_min: 196608000
+kstat.zfs.misc.arcstats.c: 1470553736
+kstat.zfs.misc.arcstats.p: 665524427
+kstat.zfs.misc.arcstats.hash_chain_max: 7
+kstat.zfs.misc.arcstats.hash_chains: 14180
+kstat.zfs.misc.arcstats.hash_collisions: 2180398
+kstat.zfs.misc.arcstats.hash_elements_max: 238188
+kstat.zfs.misc.arcstats.hash_elements: 111458
+kstat.zfs.misc.arcstats.evict_l2_ineligible: 60262400
+kstat.zfs.misc.arcstats.evict_l2_eligible: 35702978560
+kstat.zfs.misc.arcstats.evict_l2_cached: 0
+kstat.zfs.misc.arcstats.evict_skip: 21716568
+kstat.zfs.misc.arcstats.mutex_miss: 873
+kstat.zfs.misc.arcstats.recycle_miss: 5018771
+kstat.zfs.misc.arcstats.stolen: 1327563
+kstat.zfs.misc.arcstats.deleted: 1187256
+kstat.zfs.misc.arcstats.allocated: 10150518
+kstat.zfs.misc.arcstats.mfu_ghost_hits: 1408986
+kstat.zfs.misc.arcstats.mfu_hits: 51952454
+kstat.zfs.misc.arcstats.mru_ghost_hits: 696819
+kstat.zfs.misc.arcstats.mru_hits: 11115835
+kstat.zfs.misc.arcstats.prefetch_metadata_misses: 32
+kstat.zfs.misc.arcstats.prefetch_metadata_hits: 2
+kstat.zfs.misc.arcstats.prefetch_data_misses: 0
+kstat.zfs.misc.arcstats.prefetch_data_hits: 0
+kstat.zfs.misc.arcstats.demand_metadata_misses: 9231542
+kstat.zfs.misc.arcstats.demand_metadata_hits: 40650947
+kstat.zfs.misc.arcstats.demand_data_misses: 75230
+kstat.zfs.misc.arcstats.demand_data_hits: 22417340
+kstat.zfs.misc.arcstats.misses: 9306804
+kstat.zfs.misc.arcstats.hits: 63068289

collector/fixtures/zfs/zpool_stats_stdout.txt

@@ -0,0 +1,12 @@
+trout	size	4294967296	-
+trout	free	1040117248	-
+trout	allocated	70144	-
+trout	capacity	0%	-
+trout	dedupratio	1.00x	-
+trout	fragmentation	0%	-
+zroot	size	118111600640	-
+zroot	free	3990917120	-
+zroot	allocated	114120683520	-
+zroot	capacity	50%	-
+zroot	dedupratio	1.00x	-
+zroot	fragmentation	67%	-

collector/zfs.go

@@ -0,0 +1,109 @@
+package collector
+
+// +build linux freebsd
+// +build !nozfs
+
+import (
+	"errors"
+	"strings"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/common/log"
+)
+
+type zfsMetricValue int
+
+const zfsErrorValue = zfsMetricValue(-1)
+
+var zfsNotAvailableError = errors.New("ZFS / ZFS statistics are not available")
+
+type zfsSysctl string
+type zfsSubsystemName string
+
+const (
+	arc            = zfsSubsystemName("zfsArc")
+	zpoolSubsystem = zfsSubsystemName("zfsPool")
+)
+
+// Metrics
+
+type zfsMetric struct {
+	subsystem zfsSubsystemName // The Prometheus subsystem name.
+	name      string           // The Prometheus name of the metric.
+	sysctl    zfsSysctl        // The sysctl of the ZFS metric.
+}
+
+type datasetMetric struct {
+	subsystem zfsSubsystemName
+	name      string
+}
+
+// Collector
+
+func init() {
+	Factories["zfs"] = NewZFSCollector
+}
+
+type zfsCollector struct {
+	zfsMetrics []zfsMetric
+}
+
+func NewZFSCollector() (Collector, error) {
+	return &zfsCollector{}, nil
+}
+
+func (c *zfsCollector) Update(ch chan<- prometheus.Metric) (err error) {
+
+	err = c.zfsAvailable()
+	switch {
+	case err == zfsNotAvailableError:
+		log.Debug(err)
+		return nil
+	case err != nil:
+		return err
+	}
+
+	// Arcstats
+	err = c.updateArcstats(ch)
+	if err != nil {
+		return err
+	}
+
+	// Pool stats
+	return c.updatePoolStats(ch)
+}
+
+func (s zfsSysctl) metricName() string {
+	parts := strings.Split(string(s), ".")
+	return parts[len(parts)-1]
+}
+
+func (c *zfsCollector) ConstSysctlMetric(subsystem zfsSubsystemName, sysctl zfsSysctl, value zfsMetricValue) prometheus.Metric {
+
+	metricName := sysctl.metricName()
+
+	return prometheus.MustNewConstMetric(
+		prometheus.NewDesc(
+			prometheus.BuildFQName(Namespace, string(subsystem), metricName),
+			string(sysctl),
+			nil,
+			nil,
+		),
+		prometheus.UntypedValue,
+		float64(value),
+	)
+}
+
+func (c *zfsCollector) ConstZpoolMetric(pool, name string, value float64) prometheus.Metric {
+	return prometheus.MustNewConstMetric(
+		prometheus.NewDesc(
+			prometheus.BuildFQName(Namespace, string(zpoolSubsystem), name),
+			name,
+			[]string{"pool"},
+			nil,
+		),
+		prometheus.UntypedValue,
+		float64(value),
+		pool,
+	)
+}

collector/zfs_freebsd.go

@@ -0,0 +1,110 @@
+package collector
+
+import (
+	"bufio"
+	"io"
+	"os/exec"
+	"strconv"
+	"strings"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/common/log"
+)
+
+/*
+#cgo LDFLAGS:
+#include <sys/param.h>
+#include <sys/module.h>
+
+int zfsModuleLoaded() {
+	int modid = modfind("zfs");
+	return modid < 0 ? 0 : -1;
+}
+
+*/
+import "C"
+
+func (c *zfsCollector) zfsAvailable() error {
+	if C.zfsModuleLoaded() == 0 {
+		return zfsNotAvailableError
+	}
+	return nil
+}
+
+const zfsArcstatsSysctl = "kstat.zfs.misc.arcstats"
+
+func (c *zfsCollector) RunOnStdout(cmd *exec.Cmd, handler func(io.Reader) error) (err error) {
+
+	stdout, err := cmd.StdoutPipe()
+	if err != nil {
+		return
+	}
+
+	if err = cmd.Start(); err != nil {
+		return
+	}
+
+	err = handler(stdout)
+	if err != nil {
+		return
+	}
+
+	return cmd.Wait()
+}
+
+func (c *zfsCollector) updateArcstats(ch chan<- prometheus.Metric) (err error) {
+
+	cmd := exec.Command("sysctl", zfsArcstatsSysctl)
+
+	err = c.RunOnStdout(cmd, func(stdout io.Reader) error {
+		return c.parseArcstatsSysctlOutput(stdout, func(sysctl zfsSysctl, value zfsMetricValue) {
+			ch <- c.ConstSysctlMetric(arc, sysctl, zfsMetricValue(value))
+		})
+	})
+	return err
+
+}
+
+func (c *zfsCollector) parseArcstatsSysctlOutput(reader io.Reader, handler func(zfsSysctl, zfsMetricValue)) (err error) {
+
+	// Decode values
+	scanner := bufio.NewScanner(reader)
+	for scanner.Scan() {
+
+		fields := strings.Fields(scanner.Text())
+
+		if len(fields) != 2 ||
+			!strings.HasPrefix(fields[0], zfsArcstatsSysctl) ||
+			!strings.HasSuffix(fields[0], ":") {
+
+			log.Debugf("Skipping line of unknown format: %q", scanner.Text())
+			continue
+
+		}
+
+		sysctl := zfsSysctl(strings.TrimSuffix(fields[0], ":"))
+		value, err := strconv.Atoi(fields[1])
+		if err != nil {
+			return err
+		}
+
+		handler(sysctl, zfsMetricValue(value))
+	}
+	return scanner.Err()
+
+}
+
+func (c *zfsCollector) updatePoolStats(ch chan<- prometheus.Metric) (err error) {
+
+	poolProperties := []string{"size", "free", "allocated", "capacity", "dedupratio", "fragmentation"}
+
+	cmd := exec.Command("zpool", "get", "-pH", strings.Join(poolProperties, ","))
+
+	err = c.RunOnStdout(cmd, func(stdout io.Reader) error {
+		return c.parseZpoolOutput(stdout, func(pool, name string, value float64) {
+			ch <- c.ConstZpoolMetric(pool, name, value)
+		})
+	})
+
+	return err
+}

collector/zfs_freebsd_test.go

@@ -0,0 +1,44 @@
+package collector
+
+import (
+	"os"
+	"testing"
+)
+
+func TestArcstatsParsing(t *testing.T) {
+
+	arcstatsOutput, err := os.Open("fixtures/sysctl/freebsd/kstat.zfs.misc.arcstats.txt")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer arcstatsOutput.Close()
+
+	c := zfsCollector{}
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	handlerCalled := false
+	err = c.parseArcstatsSysctlOutput(arcstatsOutput, func(s zfsSysctl, v zfsMetricValue) {
+
+		if s != zfsSysctl("kstat.zfs.misc.arcstats.hits") {
+			return
+		}
+
+		handlerCalled = true
+
+		if v != zfsMetricValue(63068289) {
+			t.Fatalf("Incorrect value parsed from sysctl output")
+		}
+
+	})
+
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if !handlerCalled {
+		t.Fatal("Arcstats parsing handler was not called for some expected sysctls")
+	}
+
+}

collector/zfs_linux.go

@@ -0,0 +1,88 @@
+package collector
+
+import (
+	"bufio"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"strconv"
+	"strings"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/common/log"
+)
+
+const (
+	zfsArcstatsProcpath = "spl/kstat/zfs/arcstats"
+)
+
+func (c *zfsCollector) zfsAvailable() (err error) {
+	file, err := c.openArcstatsFile()
+	if err != nil {
+		file.Close()
+	}
+	return err
+}
+
+func (c *zfsCollector) openArcstatsFile() (file *os.File, err error) {
+	file, err = os.Open(procFilePath(zfsArcstatsProcpath))
+	if err != nil {
+		log.Debugf("Cannot open %q for reading. Is the kernel module loaded?", procFilePath(zfsArcstatsProcpath))
+		err = zfsNotAvailableError
+	}
+	return
+}
+
+func (c *zfsCollector) updateArcstats(ch chan<- prometheus.Metric) (err error) {
+
+	file, err := c.openArcstatsFile()
+	if err != nil {
+		return err
+	}
+	defer file.Close()
+
+	return c.parseArcstatsProcfsFile(file, func(s zfsSysctl, v zfsMetricValue) {
+		ch <- c.ConstSysctlMetric(arc, s, v)
+	})
+
+}
+
+func (c *zfsCollector) parseArcstatsProcfsFile(reader io.Reader, handler func(zfsSysctl, zfsMetricValue)) (err error) {
+
+	scanner := bufio.NewScanner(reader)
+
+	parseLine := false
+	for scanner.Scan() {
+
+		parts := strings.Fields(scanner.Text())
+
+		if !parseLine && len(parts) == 3 && parts[0] == "name" && parts[1] == "type" && parts[2] == "data" {
+			// Start parsing from here.
+			parseLine = true
+			continue
+		}
+
+		if !parseLine || len(parts) < 3 {
+			continue
+		}
+
+		key := fmt.Sprintf("kstat.zfs.misc.arcstats.%s", parts[0])
+
+		value, err := strconv.Atoi(parts[2])
+		if err != nil {
+			return fmt.Errorf("could not parse expected integer value for %q", key)
+		}
+		handler(zfsSysctl(key), zfsMetricValue(value))
+
+	}
+	if !parseLine {
+		return errors.New("did not parse a single arcstat metric")
+	}
+
+	return scanner.Err()
+}
+
+func (c *zfsCollector) updatePoolStats(ch chan<- prometheus.Metric) (err error) {
+	return nil
+}

collector/zfs_linux_test.go

@@ -0,0 +1,44 @@
+package collector
+
+import (
+	"os"
+	"testing"
+)
+
+func TestArcstatsParsing(t *testing.T) {
+
+	arcstatsFile, err := os.Open("fixtures/proc/spl/kstat/zfs/arcstats")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer arcstatsFile.Close()
+
+	c := zfsCollector{}
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	handlerCalled := false
+	err = c.parseArcstatsProcfsFile(arcstatsFile, func(s zfsSysctl, v zfsMetricValue) {
+
+		if s != zfsSysctl("kstat.zfs.misc.arcstats.hits") {
+			return
+		}
+
+		handlerCalled = true
+
+		if v != zfsMetricValue(8772612) {
+			t.Fatalf("Incorrect value parsed from procfs data")
+		}
+
+	})
+
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if !handlerCalled {
+		t.Fatal("Arcstats parsing handler was not called for some expected sysctls")
+	}
+
+}

collector/zfs_zpool.go

@@ -0,0 +1,41 @@
+package collector
+
+import (
+	"bufio"
+	"fmt"
+	"io"
+	"strconv"
+	"strings"
+)
+
+// zpool metrics
+
+func (c *zfsCollector) parseZpoolOutput(reader io.Reader, handler func(string, string, float64)) (err error) {
+
+	scanner := bufio.NewScanner(reader)
+	for scanner.Scan() {
+
+		fields := strings.Fields(scanner.Text())
+		if len(fields) != 4 {
+			return fmt.Errorf("Unexpected output of zpool command")
+		}
+
+		valueString := fields[2]
+		switch {
+		case strings.HasSuffix(fields[2], "%"):
+			percentage := strings.TrimSuffix(fields[2], "%")
+			valueString = "0." + percentage
+		case strings.HasSuffix(fields[2], "x"):
+			valueString = strings.TrimSuffix(fields[2], "x")
+		}
+
+		value, err := strconv.ParseFloat(valueString, 64)
+		if err != nil {
+			return err
+		}
+		handler(fields[0], fields[1], value)
+
+	}
+	return scanner.Err()
+
+}

collector/zfs_zpool_test.go

@@ -0,0 +1,59 @@
+package collector
+
+import (
+	"os"
+	"testing"
+)
+
+func TestZpoolParsing(t *testing.T) {
+
+	zpoolOutput, err := os.Open("fixtures/zfs/zpool_stats_stdout.txt")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer zpoolOutput.Close()
+
+	c := zfsCollector{}
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	pools := make([]string, 2)
+	troutSize := float64(-1)
+	troutDedupratio := float64(-1)
+	zrootCapacity := float64(-1)
+
+	err = c.parseZpoolOutput(zpoolOutput, func(pool, name string, value float64) {
+		pools = append(pools, pool)
+		if pool == "trout" && name == "size" {
+			troutSize = value
+		}
+		if pool == "trout" && name == "dedupratio" {
+			troutDedupratio = value
+		}
+		if pool == "zroot" && name == "capacity" {
+			zrootCapacity = value
+		}
+	})
+
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if pools[0] == "trout" && pools[1] == "zroot" {
+		t.Fatal("Did not parse all pools in fixture")
+	}
+
+	if troutSize != float64(4294967296) {
+		t.Fatal("Unexpected value for pool 'trout's size value")
+	}
+
+	if troutDedupratio != float64(1.0) {
+		t.Fatal("Unexpected value for pool 'trout's dedupratio value")
+	}
+
+	if zrootCapacity != float64(0.5) {
+		t.Fatal("Unexpected value for pool 'zroot's capacity value")
+	}
+
+}