feat: add prebuild timing metrics to Prometheus (#19503)

## Description

This PR introduces one counter and two histograms related to workspace
creation and claiming. The goal is to provide clearer observability into
how workspaces are created (regular vs prebuild) and the time cost of
those operations.

### `coderd_workspace_creation_total`

* Metric type: Counter
* Name: `coderd_workspace_creation_total`
* Labels: `organization_name`, `template_name`, `preset_name`

This counter tracks whether a regular workspace (not created from a
prebuild pool) was created using a preset or not.
Currently, we already expose `coderd_prebuilt_workspaces_claimed_total`
for claimed prebuilt workspaces, but we lack a comparable metric for
regular workspace creations. This metric fills that gap, making it
possible to compare regular creations against claims.

Implementation notes:
* Exposed as a `coderd_` metric, consistent with other workspace-related
metrics (e.g. `coderd_api_workspace_latest_build`:
https://github.com/coder/coder/blob/main/coderd/prometheusmetrics/prometheusmetrics.go#L149).
* Every `defaultRefreshRate` (1 minute ), DB query
`GetRegularWorkspaceCreateMetrics` is executed to fetch all regular
workspaces (not created from a prebuild pool).
* The counter is updated with the total from all time (not just since
metric introduction). This differs from the histograms below, which only
accumulate from their introduction forward.

### `coderd_workspace_creation_duration_seconds` &
`coderd_prebuilt_workspace_claim_duration_seconds`

* Metric types: Histogram
* Names:
  * `coderd_workspace_creation_duration_seconds`
* Labels: `organization_name`, `template_name`, `preset_name`, `type`
(`regular`, `prebuild`)
  * `coderd_prebuilt_workspace_claim_duration_seconds`
    * Labels: `organization_name`, `template_name`, `preset_name`

We already have `coderd_provisionerd_workspace_build_timings_seconds`,
which tracks build run times for all workspace builds handled by the
provisioner daemon.
However, in the context of this issue, we are only interested in
creation and claim build times, not all transitions; additionally, this
metric does not include `preset_name`, and adding it there would
significantly increase cardinality. Therefore, separate more focused
metrics are introduced here:
* `coderd_workspace_creation_duration_seconds`: Build time to create a
workspace (either a regular workspace or the build into a prebuild pool,
for prebuild initial provisioning build).
* `coderd_prebuilt_workspace_claim_duration_seconds`: Time to claim a
prebuilt workspace from the pool.

The reason for two separate histograms is that:
* Creation (regular or prebuild): provisioning builds with similar time
magnitude, generally expected to take longer than a claim operation.
* Claim: expected to be a much faster provisioning build.

#### Native histogram usage

Provisioning times vary widely between projects. Using static buckets
risks unbalanced or poorly informative histograms.
To address this, these metrics use [Prometheus native
histograms](https://prometheus.io/docs/specs/native_histograms/):
* First introduced in Prometheus v2.40.0
* Recommended stable usage from v2.45+
* Requires Go client `prometheus/client_golang` v1.15.0+
* Experimental and must be explicitly enabled on the server
(`--enable-feature=native-histograms`)

For compatibility, we also retain a classic bucket definition (aligned
with the existing provisioner metric:
https://github.com/coder/coder/blob/main/provisionerd/provisionerd.go#L182-L189).
* If native histograms are enabled, Prometheus ingests the
high-resolution histogram.
* If not, it falls back to the predefined buckets.

Implementation notes:
* Unlike the counter, these histograms are updated in real-time at
workspace build job completion.
* They reflect data only from the point of introduction forward (no
historical backfill).

## Relates to 

Closes: https://github.com/coder/coder/issues/19528
Native histograms tested in observability stack:
https://github.com/coder/observability/pull/50

coderd/coderd.go

@@ -241,6 +241,8 @@ type Options struct {
 	UpdateAgentMetrics func(ctx context.Context, labels prometheusmetrics.AgentMetricLabels, metrics []*agentproto.Stats_Metric)
 	StatsBatcher       workspacestats.Batcher
 
+	ProvisionerdServerMetrics *provisionerdserver.Metrics
+
 	// WorkspaceAppAuditSessionTimeout allows changing the timeout for audit
 	// sessions. Raising or lowering this value will directly affect the write
 	// load of the audit log table. This is used for testing. Default 1 hour.
@@ -1930,6 +1932,7 @@ func (api *API) CreateInMemoryTaggedProvisionerDaemon(dialCtx context.Context, n
 		},
 		api.NotificationsEnqueuer,
 		&api.PrebuildsReconciler,
+		api.ProvisionerdServerMetrics,
 	)
 	if err != nil {
 		return nil, err

coderd/coderdtest/coderdtest.go

@@ -184,6 +184,8 @@ type Options struct {
 	OIDCConvertKeyCache                cryptokeys.SigningKeycache
 	Clock                              quartz.Clock
 	TelemetryReporter                  telemetry.Reporter
+
+	ProvisionerdServerMetrics *provisionerdserver.Metrics
 }
 
 // New constructs a codersdk client connected to an in-memory API instance.
@@ -604,6 +606,7 @@ func NewOptions(t testing.TB, options *Options) (func(http.Handler), context.Can
 			Clock:                              options.Clock,
 			AppEncryptionKeyCache:              options.APIKeyEncryptionCache,
 			OIDCConvertKeyCache:                options.OIDCConvertKeyCache,
+			ProvisionerdServerMetrics:          options.ProvisionerdServerMetrics,
 		}
 }
 

coderd/database/dbauthz/dbauthz.go

@@ -2699,6 +2699,13 @@ func (q *querier) GetQuotaConsumedForUser(ctx context.Context, params database.G
 	return q.db.GetQuotaConsumedForUser(ctx, params)
 }
 
+func (q *querier) GetRegularWorkspaceCreateMetrics(ctx context.Context) ([]database.GetRegularWorkspaceCreateMetricsRow, error) {
+	if err := q.authorizeContext(ctx, policy.ActionRead, rbac.ResourceWorkspace.All()); err != nil {
+		return nil, err
+	}
+	return q.db.GetRegularWorkspaceCreateMetrics(ctx)
+}
+
 func (q *querier) GetReplicaByID(ctx context.Context, id uuid.UUID) (database.Replica, error) {
 	if err := q.authorizeContext(ctx, policy.ActionRead, rbac.ResourceSystem); err != nil {
 		return database.Replica{}, err

coderd/database/dbauthz/dbauthz_test.go

@@ -2177,6 +2177,10 @@ func (s *MethodTestSuite) TestWorkspace() {
 		dbm.EXPECT().GetWorkspaceAgentDevcontainersByAgentID(gomock.Any(), agt.ID).Return([]database.WorkspaceAgentDevcontainer{d}, nil).AnyTimes()
 		check.Args(agt.ID).Asserts(w, policy.ActionRead).Returns([]database.WorkspaceAgentDevcontainer{d})
 	}))
+	s.Run("GetRegularWorkspaceCreateMetrics", s.Subtest(func(_ database.Store, check *expects) {
+		check.Args().
+			Asserts(rbac.ResourceWorkspace.All(), policy.ActionRead)
+	}))
 }
 
 func (s *MethodTestSuite) TestWorkspacePortSharing() {

coderd/database/dbmetrics/querymetrics.go

@@ -1356,6 +1356,13 @@ func (m queryMetricsStore) GetQuotaConsumedForUser(ctx context.Context, ownerID
 	return consumed, err
 }
 
+func (m queryMetricsStore) GetRegularWorkspaceCreateMetrics(ctx context.Context) ([]database.GetRegularWorkspaceCreateMetricsRow, error) {
+	start := time.Now()
+	r0, r1 := m.s.GetRegularWorkspaceCreateMetrics(ctx)
+	m.queryLatencies.WithLabelValues("GetRegularWorkspaceCreateMetrics").Observe(time.Since(start).Seconds())
+	return r0, r1
+}
+
 func (m queryMetricsStore) GetReplicaByID(ctx context.Context, id uuid.UUID) (database.Replica, error) {
 	start := time.Now()
 	replica, err := m.s.GetReplicaByID(ctx, id)

coderd/database/dbmock/dbmock.go

@@ -2851,6 +2851,21 @@ func (mr *MockStoreMockRecorder) GetQuotaConsumedForUser(ctx, arg any) *gomock.C
 	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetQuotaConsumedForUser", reflect.TypeOf((*MockStore)(nil).GetQuotaConsumedForUser), ctx, arg)
 }
 
+// GetRegularWorkspaceCreateMetrics mocks base method.
+func (m *MockStore) GetRegularWorkspaceCreateMetrics(ctx context.Context) ([]database.GetRegularWorkspaceCreateMetricsRow, error) {
+	m.ctrl.T.Helper()
+	ret := m.ctrl.Call(m, "GetRegularWorkspaceCreateMetrics", ctx)
+	ret0, _ := ret[0].([]database.GetRegularWorkspaceCreateMetricsRow)
+	ret1, _ := ret[1].(error)
+	return ret0, ret1
+}
+
+// GetRegularWorkspaceCreateMetrics indicates an expected call of GetRegularWorkspaceCreateMetrics.
+func (mr *MockStoreMockRecorder) GetRegularWorkspaceCreateMetrics(ctx any) *gomock.Call {
+	mr.mock.ctrl.T.Helper()
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetRegularWorkspaceCreateMetrics", reflect.TypeOf((*MockStore)(nil).GetRegularWorkspaceCreateMetrics), ctx)
+}
+
 // GetReplicaByID mocks base method.
 func (m *MockStore) GetReplicaByID(ctx context.Context, id uuid.UUID) (database.Replica, error) {
 	m.ctrl.T.Helper()

coderd/database/querier.go

@@ -306,6 +306,9 @@ type sqlcQuerier interface {
 	GetProvisionerLogsAfterID(ctx context.Context, arg GetProvisionerLogsAfterIDParams) ([]ProvisionerJobLog, error)
 	GetQuotaAllowanceForUser(ctx context.Context, arg GetQuotaAllowanceForUserParams) (int64, error)
 	GetQuotaConsumedForUser(ctx context.Context, arg GetQuotaConsumedForUserParams) (int64, error)
+	// Count regular workspaces: only those whose first successful 'start' build
+	// was not initiated by the prebuild system user.
+	GetRegularWorkspaceCreateMetrics(ctx context.Context) ([]GetRegularWorkspaceCreateMetricsRow, error)
 	GetReplicaByID(ctx context.Context, id uuid.UUID) (Replica, error)
 	GetReplicasUpdatedAfter(ctx context.Context, updatedAt time.Time) ([]Replica, error)
 	GetRunningPrebuiltWorkspaces(ctx context.Context) ([]GetRunningPrebuiltWorkspacesRow, error)

coderd/database/queries.sql.go

@@ -7309,7 +7309,7 @@ const getPrebuildMetrics = `-- name: GetPrebuildMetrics :many
 SELECT
 	t.name as template_name,
 	tvp.name as preset_name,
-		o.name as organization_name,
+	o.name as organization_name,
 	COUNT(*) as created_count,
 	COUNT(*) FILTER (WHERE pj.job_status = 'failed'::provisioner_job_status) as failed_count,
 	COUNT(*) FILTER (
@@ -20131,6 +20131,75 @@ func (q *sqlQuerier) GetDeploymentWorkspaceStats(ctx context.Context) (GetDeploy
 	return i, err
 }
 
+const getRegularWorkspaceCreateMetrics = `-- name: GetRegularWorkspaceCreateMetrics :many
+WITH first_success_build AS (
+	-- Earliest successful 'start' build per workspace
+	SELECT DISTINCT ON (wb.workspace_id)
+		wb.workspace_id,
+		wb.template_version_preset_id,
+		wb.initiator_id
+	FROM workspace_builds wb
+	JOIN provisioner_jobs pj ON pj.id = wb.job_id
+	WHERE
+		wb.transition = 'start'::workspace_transition
+  		AND pj.job_status = 'succeeded'::provisioner_job_status
+	ORDER BY wb.workspace_id, wb.build_number, wb.id
+)
+SELECT
+	t.name AS template_name,
+	COALESCE(tvp.name, '') AS preset_name,
+	o.name AS organization_name,
+	COUNT(*) AS created_count
+FROM first_success_build fsb
+	JOIN workspaces w ON w.id = fsb.workspace_id
+	JOIN templates t ON t.id = w.template_id
+	LEFT JOIN template_version_presets tvp ON tvp.id = fsb.template_version_preset_id
+	JOIN organizations o ON o.id = w.organization_id
+WHERE
+	NOT t.deleted
+	-- Exclude workspaces whose first successful start was the prebuilds system user
+	AND fsb.initiator_id != 'c42fdf75-3097-471c-8c33-fb52454d81c0'::uuid
+GROUP BY t.name, COALESCE(tvp.name, ''), o.name
+ORDER BY t.name, preset_name, o.name
+`
+
+type GetRegularWorkspaceCreateMetricsRow struct {
+	TemplateName     string `db:"template_name" json:"template_name"`
+	PresetName       string `db:"preset_name" json:"preset_name"`
+	OrganizationName string `db:"organization_name" json:"organization_name"`
+	CreatedCount     int64  `db:"created_count" json:"created_count"`
+}
+
+// Count regular workspaces: only those whose first successful 'start' build
+// was not initiated by the prebuild system user.
+func (q *sqlQuerier) GetRegularWorkspaceCreateMetrics(ctx context.Context) ([]GetRegularWorkspaceCreateMetricsRow, error) {
+	rows, err := q.db.QueryContext(ctx, getRegularWorkspaceCreateMetrics)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var items []GetRegularWorkspaceCreateMetricsRow
+	for rows.Next() {
+		var i GetRegularWorkspaceCreateMetricsRow
+		if err := rows.Scan(
+			&i.TemplateName,
+			&i.PresetName,
+			&i.OrganizationName,
+			&i.CreatedCount,
+		); err != nil {
+			return nil, err
+		}
+		items = append(items, i)
+	}
+	if err := rows.Close(); err != nil {
+		return nil, err
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+	return items, nil
+}
+
 const getWorkspaceACLByID = `-- name: GetWorkspaceACLByID :one
 SELECT
 	group_acl as groups,

coderd/database/queries/prebuilds.sql

@@ -230,7 +230,7 @@ HAVING COUNT(*) = @hard_limit::bigint;
 SELECT
 	t.name as template_name,
 	tvp.name as preset_name,
-		o.name as organization_name,
+	o.name as organization_name,
 	COUNT(*) as created_count,
 	COUNT(*) FILTER (WHERE pj.job_status = 'failed'::provisioner_job_status) as failed_count,
 	COUNT(*) FILTER (

coderd/database/queries/workspaces.sql

@@ -923,3 +923,36 @@ SET
 	user_acl = @user_acl
 WHERE
 	id = @id;
+
+-- name: GetRegularWorkspaceCreateMetrics :many
+-- Count regular workspaces: only those whose first successful 'start' build
+-- was not initiated by the prebuild system user.
+WITH first_success_build AS (
+	-- Earliest successful 'start' build per workspace
+	SELECT DISTINCT ON (wb.workspace_id)
+		wb.workspace_id,
+		wb.template_version_preset_id,
+		wb.initiator_id
+	FROM workspace_builds wb
+	JOIN provisioner_jobs pj ON pj.id = wb.job_id
+	WHERE
+		wb.transition = 'start'::workspace_transition
+  		AND pj.job_status = 'succeeded'::provisioner_job_status
+	ORDER BY wb.workspace_id, wb.build_number, wb.id
+)
+SELECT
+	t.name AS template_name,
+	COALESCE(tvp.name, '') AS preset_name,
+	o.name AS organization_name,
+	COUNT(*) AS created_count
+FROM first_success_build fsb
+	JOIN workspaces w ON w.id = fsb.workspace_id
+	JOIN templates t ON t.id = w.template_id
+	LEFT JOIN template_version_presets tvp ON tvp.id = fsb.template_version_preset_id
+	JOIN organizations o ON o.id = w.organization_id
+WHERE
+	NOT t.deleted
+	-- Exclude workspaces whose first successful start was the prebuilds system user
+	AND fsb.initiator_id != 'c42fdf75-3097-471c-8c33-fb52454d81c0'::uuid
+GROUP BY t.name, COALESCE(tvp.name, ''), o.name
+ORDER BY t.name, preset_name, o.name;

coderd/prometheusmetrics/prometheusmetrics.go

@@ -165,6 +165,18 @@ func Workspaces(ctx context.Context, logger slog.Logger, registerer prometheus.R
 		return nil, err
 	}
 
+	workspaceCreationTotal := prometheus.NewCounterVec(
+		prometheus.CounterOpts{
+			Namespace: "coderd",
+			Name:      "workspace_creation_total",
+			Help:      "Total regular (non-prebuilt) workspace creations by organization, template, and preset.",
+		},
+		[]string{"organization_name", "template_name", "preset_name"},
+	)
+	if err := registerer.Register(workspaceCreationTotal); err != nil {
+		return nil, err
+	}
+
 	ctx, cancelFunc := context.WithCancel(ctx)
 	done := make(chan struct{})
 
@@ -200,6 +212,27 @@ func Workspaces(ctx context.Context, logger slog.Logger, registerer prometheus.R
 				string(w.LatestBuildTransition),
 			).Add(1)
 		}
+
+		// Update regular workspaces (without a prebuild transition) creation counter
+		regularWorkspaces, err := db.GetRegularWorkspaceCreateMetrics(ctx)
+		if err != nil {
+			if errors.Is(err, sql.ErrNoRows) {
+				workspaceCreationTotal.Reset()
+			} else {
+				logger.Warn(ctx, "failed to load regular workspaces for metrics", slog.Error(err))
+			}
+			return
+		}
+
+		workspaceCreationTotal.Reset()
+
+		for _, regularWorkspace := range regularWorkspaces {
+			workspaceCreationTotal.WithLabelValues(
+				regularWorkspace.OrganizationName,
+				regularWorkspace.TemplateName,
+				regularWorkspace.PresetName,
+			).Add(float64(regularWorkspace.CreatedCount))
+		}
 	}
 
 	// Use time.Nanosecond to force an initial tick. It will be reset to the

coderd/prometheusmetrics/prometheusmetrics_test.go

@@ -424,6 +424,107 @@ func TestWorkspaceLatestBuildStatuses(t *testing.T) {
 	}
 }
 
+func TestWorkspaceCreationTotal(t *testing.T) {
+	t.Parallel()
+
+	for _, tc := range []struct {
+		Name               string
+		Database           func() database.Store
+		ExpectedWorkspaces int
+	}{
+		{
+			Name: "None",
+			Database: func() database.Store {
+				db, _ := dbtestutil.NewDB(t)
+				return db
+			},
+			ExpectedWorkspaces: 0,
+		},
+		{
+			// Should count only the successfully created workspaces
+			Name: "Multiple",
+			Database: func() database.Store {
+				db, _ := dbtestutil.NewDB(t)
+				u := dbgen.User(t, db, database.User{})
+				org := dbgen.Organization(t, db, database.Organization{})
+				insertTemplates(t, db, u, org)
+				insertCanceled(t, db, u, org)
+				insertFailed(t, db, u, org)
+				insertFailed(t, db, u, org)
+				insertSuccess(t, db, u, org)
+				insertSuccess(t, db, u, org)
+				insertSuccess(t, db, u, org)
+				insertRunning(t, db, u, org)
+				return db
+			},
+			ExpectedWorkspaces: 3,
+		},
+		{
+			// Should not include prebuilt workspaces
+			Name: "MultipleWithPrebuild",
+			Database: func() database.Store {
+				ctx := context.Background()
+				db, _ := dbtestutil.NewDB(t)
+				u := dbgen.User(t, db, database.User{})
+				prebuildUser, err := db.GetUserByID(ctx, database.PrebuildsSystemUserID)
+				require.NoError(t, err)
+				org := dbgen.Organization(t, db, database.Organization{})
+				insertTemplates(t, db, u, org)
+				insertCanceled(t, db, u, org)
+				insertFailed(t, db, u, org)
+				insertSuccess(t, db, u, org)
+				insertSuccess(t, db, prebuildUser, org)
+				insertRunning(t, db, u, org)
+				return db
+			},
+			ExpectedWorkspaces: 1,
+		},
+		{
+			// Should include deleted workspaces
+			Name: "MultipleWithDeleted",
+			Database: func() database.Store {
+				db, _ := dbtestutil.NewDB(t)
+				u := dbgen.User(t, db, database.User{})
+				org := dbgen.Organization(t, db, database.Organization{})
+				insertTemplates(t, db, u, org)
+				insertCanceled(t, db, u, org)
+				insertFailed(t, db, u, org)
+				insertSuccess(t, db, u, org)
+				insertRunning(t, db, u, org)
+				insertDeleted(t, db, u, org)
+				return db
+			},
+			ExpectedWorkspaces: 2,
+		},
+	} {
+		t.Run(tc.Name, func(t *testing.T) {
+			t.Parallel()
+			registry := prometheus.NewRegistry()
+			closeFunc, err := prometheusmetrics.Workspaces(context.Background(), testutil.Logger(t), registry, tc.Database(), testutil.IntervalFast)
+			require.NoError(t, err)
+			t.Cleanup(closeFunc)
+
+			require.Eventually(t, func() bool {
+				metrics, err := registry.Gather()
+				assert.NoError(t, err)
+
+				sum := 0
+				for _, m := range metrics {
+					if m.GetName() != "coderd_workspace_creation_total" {
+						continue
+					}
+					for _, metric := range m.Metric {
+						sum += int(metric.GetCounter().GetValue())
+					}
+				}
+
+				t.Logf("count = %d, expected == %d", sum, tc.ExpectedWorkspaces)
+				return sum == tc.ExpectedWorkspaces
+			}, testutil.WaitShort, testutil.IntervalFast)
+		})
+	}
+}
+
 func TestAgents(t *testing.T) {
 	t.Parallel()
 
@@ -897,6 +998,7 @@ func insertRunning(t *testing.T, db database.Store, u database.User, org databas
 		Transition:        database.WorkspaceTransitionStart,
 		Reason:            database.BuildReasonInitiator,
 		TemplateVersionID: templateVersionID,
+		InitiatorID:       u.ID,
 	})
 	require.NoError(t, err)
 	// This marks the job as started.

coderd/provisionerdserver/metrics.go

@@ -0,0 +1,177 @@
+package provisionerdserver
+
+import (
+	"context"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
+
+	"cdr.dev/slog"
+)
+
+type Metrics struct {
+	logger                   slog.Logger
+	workspaceCreationTimings *prometheus.HistogramVec
+	workspaceClaimTimings    *prometheus.HistogramVec
+}
+
+type WorkspaceTimingType int
+
+const (
+	Unsupported WorkspaceTimingType = iota
+	WorkspaceCreation
+	PrebuildCreation
+	PrebuildClaim
+)
+
+const (
+	workspaceTypeRegular  = "regular"
+	workspaceTypePrebuild = "prebuild"
+)
+
+type WorkspaceTimingFlags struct {
+	IsPrebuild   bool
+	IsClaim      bool
+	IsFirstBuild bool
+}
+
+func NewMetrics(logger slog.Logger) *Metrics {
+	log := logger.Named("provisionerd_server_metrics")
+
+	return &Metrics{
+		logger: log,
+		workspaceCreationTimings: prometheus.NewHistogramVec(prometheus.HistogramOpts{
+			Namespace: "coderd",
+			Name:      "workspace_creation_duration_seconds",
+			Help:      "Time to create a workspace by organization, template, preset, and type (regular or prebuild).",
+			Buckets: []float64{
+				1, // 1s
+				10,
+				30,
+				60, // 1min
+				60 * 5,
+				60 * 10,
+				60 * 30, // 30min
+				60 * 60, // 1hr
+			},
+			NativeHistogramBucketFactor: 1.1,
+			// Max number of native buckets kept at once to bound memory.
+			NativeHistogramMaxBucketNumber: 100,
+			// Merge/flush small buckets periodically to control churn.
+			NativeHistogramMinResetDuration: time.Hour,
+			// Treat tiny values as zero (helps with noisy near-zero latencies).
+			NativeHistogramZeroThreshold:    0,
+			NativeHistogramMaxZeroThreshold: 0,
+		}, []string{"organization_name", "template_name", "preset_name", "type"}),
+		workspaceClaimTimings: prometheus.NewHistogramVec(prometheus.HistogramOpts{
+			Namespace: "coderd",
+			Name:      "prebuilt_workspace_claim_duration_seconds",
+			Help:      "Time to claim a prebuilt workspace by organization, template, and preset.",
+			// Higher resolution between 1–5m to show typical prebuild claim times.
+			// Cap at 5m since longer claims diminish prebuild value.
+			Buckets: []float64{
+				1, // 1s
+				5,
+				10,
+				20,
+				30,
+				60,  // 1m
+				120, // 2m
+				180, // 3m
+				240, // 4m
+				300, // 5m
+			},
+			NativeHistogramBucketFactor: 1.1,
+			// Max number of native buckets kept at once to bound memory.
+			NativeHistogramMaxBucketNumber: 100,
+			// Merge/flush small buckets periodically to control churn.
+			NativeHistogramMinResetDuration: time.Hour,
+			// Treat tiny values as zero (helps with noisy near-zero latencies).
+			NativeHistogramZeroThreshold:    0,
+			NativeHistogramMaxZeroThreshold: 0,
+		}, []string{"organization_name", "template_name", "preset_name"}),
+	}
+}
+
+func (m *Metrics) Register(reg prometheus.Registerer) error {
+	if err := reg.Register(m.workspaceCreationTimings); err != nil {
+		return err
+	}
+	return reg.Register(m.workspaceClaimTimings)
+}
+
+func (f WorkspaceTimingFlags) count() int {
+	count := 0
+	if f.IsPrebuild {
+		count++
+	}
+	if f.IsClaim {
+		count++
+	}
+	if f.IsFirstBuild {
+		count++
+	}
+	return count
+}
+
+// getWorkspaceTimingType returns the type of the workspace build:
+//   - isPrebuild: if the workspace build corresponds to the creation of a prebuilt workspace
+//   - isClaim: if the workspace build corresponds to the claim of a prebuilt workspace
+//   - isWorkspaceFirstBuild: if the workspace build corresponds to the creation of a regular workspace
+//     (not created from the prebuild pool)
+func getWorkspaceTimingType(flags WorkspaceTimingFlags) WorkspaceTimingType {
+	switch {
+	case flags.IsPrebuild:
+		return PrebuildCreation
+	case flags.IsClaim:
+		return PrebuildClaim
+	case flags.IsFirstBuild:
+		return WorkspaceCreation
+	default:
+		return Unsupported
+	}
+}
+
+// UpdateWorkspaceTimingsMetrics updates the workspace timing metrics based on the workspace build type
+func (m *Metrics) UpdateWorkspaceTimingsMetrics(
+	ctx context.Context,
+	flags WorkspaceTimingFlags,
+	organizationName string,
+	templateName string,
+	presetName string,
+	buildTime float64,
+) {
+	m.logger.Debug(ctx, "update workspace timings metrics",
+		"organizationName", organizationName,
+		"templateName", templateName,
+		"presetName", presetName,
+		"isPrebuild", flags.IsPrebuild,
+		"isClaim", flags.IsClaim,
+		"isWorkspaceFirstBuild", flags.IsFirstBuild)
+
+	if flags.count() > 1 {
+		m.logger.Warn(ctx, "invalid workspace timing flags",
+			"isPrebuild", flags.IsPrebuild,
+			"isClaim", flags.IsClaim,
+			"isWorkspaceFirstBuild", flags.IsFirstBuild)
+		return
+	}
+
+	workspaceTimingType := getWorkspaceTimingType(flags)
+	switch workspaceTimingType {
+	case WorkspaceCreation:
+		// Regular workspace creation (without prebuild pool)
+		m.workspaceCreationTimings.
+			WithLabelValues(organizationName, templateName, presetName, workspaceTypeRegular).Observe(buildTime)
+	case PrebuildCreation:
+		// Prebuilt workspace creation duration
+		m.workspaceCreationTimings.
+			WithLabelValues(organizationName, templateName, presetName, workspaceTypePrebuild).Observe(buildTime)
+	case PrebuildClaim:
+		// Prebuilt workspace claim duration
+		m.workspaceClaimTimings.
+			WithLabelValues(organizationName, templateName, presetName).Observe(buildTime)
+	default:
+		m.logger.Warn(ctx, "unsupported workspace timing flags")
+	}
+}

coderd/provisionerdserver/provisionerdserver.go

@@ -129,6 +129,8 @@ type server struct {
 
 	heartbeatInterval time.Duration
 	heartbeatFn       func(ctx context.Context) error
+
+	metrics *Metrics
 }
 
 // We use the null byte (0x00) in generating a canonical map key for tags, so
@@ -178,6 +180,7 @@ func NewServer(
 	options Options,
 	enqueuer notifications.Enqueuer,
 	prebuildsOrchestrator *atomic.Pointer[prebuilds.ReconciliationOrchestrator],
+	metrics *Metrics,
 ) (proto.DRPCProvisionerDaemonServer, error) {
 	// Fail-fast if pointers are nil
 	if lifecycleCtx == nil {
@@ -248,6 +251,7 @@ func NewServer(
 		heartbeatFn:                 options.HeartbeatFn,
 		PrebuildsOrchestrator:       prebuildsOrchestrator,
 		UsageInserter:               usageInserter,
+		metrics:                     metrics,
 	}
 
 	if s.heartbeatFn == nil {
@@ -2281,6 +2285,50 @@ func (s *server) completeWorkspaceBuildJob(ctx context.Context, job database.Pro
 		}
 	}
 
+	// Update workspace (regular and prebuild) timing metrics
+	if s.metrics != nil {
+		// Only consider 'start' workspace builds
+		if workspaceBuild.Transition == database.WorkspaceTransitionStart {
+			// Get the updated job to report the metrics with correct data
+			updatedJob, err := s.Database.GetProvisionerJobByID(ctx, jobID)
+			if err != nil {
+				s.Logger.Error(ctx, "get updated job from database", slog.Error(err))
+			} else
+			// Only consider 'succeeded' provisioner jobs
+			if updatedJob.JobStatus == database.ProvisionerJobStatusSucceeded {
+				presetName := ""
+				if workspaceBuild.TemplateVersionPresetID.Valid {
+					preset, err := s.Database.GetPresetByID(ctx, workspaceBuild.TemplateVersionPresetID.UUID)
+					if err != nil {
+						if !errors.Is(err, sql.ErrNoRows) {
+							s.Logger.Error(ctx, "get preset by ID for workspace timing metrics", slog.Error(err))
+						}
+					} else {
+						presetName = preset.Name
+					}
+				}
+
+				buildTime := updatedJob.CompletedAt.Time.Sub(updatedJob.StartedAt.Time).Seconds()
+				s.metrics.UpdateWorkspaceTimingsMetrics(
+					ctx,
+					WorkspaceTimingFlags{
+						// Is a prebuilt workspace creation build
+						IsPrebuild: input.PrebuiltWorkspaceBuildStage.IsPrebuild(),
+						// Is a prebuilt workspace claim build
+						IsClaim: input.PrebuiltWorkspaceBuildStage.IsPrebuiltWorkspaceClaim(),
+						// Is a regular workspace creation build
+						// Only consider the first build number for regular workspaces
+						IsFirstBuild: workspaceBuild.BuildNumber == 1,
+					},
+					workspace.OrganizationName,
+					workspace.TemplateName,
+					presetName,
+					buildTime,
+				)
+			}
+		}
+	}
+
 	msg, err := json.Marshal(wspubsub.WorkspaceEvent{
 		Kind:        wspubsub.WorkspaceEventKindStateChange,
 		WorkspaceID: workspace.ID,

coderd/provisionerdserver/provisionerdserver_test.go

@@ -4144,6 +4144,7 @@ func setup(t *testing.T, ignoreLogErrors bool, ov *overrides) (proto.DRPCProvisi
 		},
 		notifEnq,
 		&op,
+		provisionerdserver.NewMetrics(logger),
 	)
 	require.NoError(t, err)
 	return srv, db, ps, daemon