Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 36 additions & 6 deletions charts/gha-runner-scale-set-experimental/values.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -323,15 +323,33 @@ listener:
## metrics configuration for the listener.
## In order to avoid helm merging these fields, we left the metrics commented out.
## When configuring metrics, please uncomment the metrics object below.
## You can modify the configuration to remove the label or specify custom buckets for histogram.
## You can modify the configuration to add or remove labels or specify custom buckets for histograms.
##
## If the buckets field is not specified, the default buckets will be applied. Default buckets are
## provided here for documentation purposes
## The example below lists every metric and every label the listener can export, together with the
## default buckets. It is not identical to the built-in defaults used when listener.metrics is not set: the
## labels marked "not exposed by default" ("job_workflow_ref", "job_workflow_name" and
## "job_workflow_target") are opt-in, so remove them to keep the default label set. If the buckets
## field is not specified, the default buckets will be applied.
##
## "job_workflow_ref" and "job_workflow_target" contain the git ref the workflow ran from (for example
## refs/heads/<branch> or refs/pull/<number>/merge), so every branch and pull request creates a new
## time series per job. On histograms this is multiplied by the number of buckets. Only keep these
## labels when your metrics backend can handle the resulting cardinality.
# metrics:
# counters:
# gha_started_jobs_total:
# labels:
# ["repository", "organization", "enterprise", "job_name", "event_name", "job_workflow_ref", "job_workflow_name", "job_workflow_target"]
# [
# "repository",
# "organization",
# "enterprise",
# "job_name",
# "event_name",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target",
# ]
# gha_completed_jobs_total:
# labels:
# [
Expand All @@ -341,6 +359,7 @@ listener:
# "job_name",
# "event_name",
# "job_result",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target",
Expand All @@ -365,7 +384,17 @@ listener:
# histograms:
# gha_job_startup_duration_seconds:
# labels:
# ["repository", "organization", "enterprise", "job_name", "event_name","job_workflow_ref", "job_workflow_name", "job_workflow_target"]

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hey, this is a good suggestion, but I'd prefer if we include all and document which ones are not exposed by default. Can you do that please?

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@nikola-jokic done in a9e2b43: both charts keep all labels in the example, with the three job_workflow_* labels marked # not exposed by default: inside each label list and the header rewritten to say the example is the full set, not the defaults. The test now checks the example lists every label the exporter populates and that the labels beyond defaultMetrics are exactly those three. Could you take another look?

# [
# "repository",
# "organization",
# "enterprise",
# "job_name",
# "event_name",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target",
# ]
# buckets:
# [
# 0.01,
Expand Down Expand Up @@ -423,9 +452,10 @@ listener:
# "job_name",
# "event_name",
# "job_result",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target"
# "job_workflow_target",
# ]
# buckets:
# [
Expand Down
42 changes: 36 additions & 6 deletions charts/gha-runner-scale-set/values.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -155,15 +155,33 @@ listenerConfig:
## listenerMetrics are configurable metrics applied to the listener.
## In order to avoid helm merging these fields, we left the metrics commented out.
## When configuring metrics, please uncomment the listenerMetrics object below.
## You can modify the configuration to remove the label or specify custom buckets for histogram.
## You can modify the configuration to add or remove labels or specify custom buckets for histograms.
##
## If the buckets field is not specified, the default buckets will be applied. Default buckets are
## provided here for documentation purposes
## The example below lists every metric and every label the listener can export, together with the
## default buckets. It is not identical to the built-in defaults used when listenerMetrics is not set: the
## labels marked "not exposed by default" ("job_workflow_ref", "job_workflow_name" and
## "job_workflow_target") are opt-in, so remove them to keep the default label set. If the buckets
## field is not specified, the default buckets will be applied.
##
## "job_workflow_ref" and "job_workflow_target" contain the git ref the workflow ran from (for example
## refs/heads/<branch> or refs/pull/<number>/merge), so every branch and pull request creates a new
## time series per job. On histograms this is multiplied by the number of buckets. Only keep these
## labels when your metrics backend can handle the resulting cardinality.
# listenerMetrics:
# counters:
# gha_started_jobs_total:
# labels:
# ["repository", "organization", "enterprise", "job_name", "event_name", "job_workflow_ref", "job_workflow_name", "job_workflow_target"]
# [
# "repository",
# "organization",
# "enterprise",
# "job_name",
# "event_name",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target",
# ]
# gha_completed_jobs_total:
# labels:
# [
Expand All @@ -173,6 +191,7 @@ listenerConfig:
# "job_name",
# "event_name",
# "job_result",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target",
Expand All @@ -197,7 +216,17 @@ listenerConfig:
# histograms:
# gha_job_startup_duration_seconds:
# labels:
# ["repository", "organization", "enterprise", "job_name", "event_name","job_workflow_ref", "job_workflow_name", "job_workflow_target"]
# [
# "repository",
# "organization",
# "enterprise",
# "job_name",
# "event_name",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target",
# ]
# buckets:
# [
# 0.01,
Expand Down Expand Up @@ -255,9 +284,10 @@ listenerConfig:
# "job_name",
# "event_name",
# "job_result",
# # not exposed by default:
# "job_workflow_ref",
# "job_workflow_name",
# "job_workflow_target"
# "job_workflow_target",
# ]
# buckets:
# [
Expand Down
146 changes: 146 additions & 0 deletions cmd/ghalistener/metrics/chart_values_test.go
Original file line number Diff line number Diff line change
@@ -0,0 +1,146 @@
package metrics

import (
"bufio"
"os"
"path/filepath"
"slices"
"strings"
"testing"

"github.com/actions/actions-runner-controller/apis/actions.github.com/v1alpha1"
"github.com/actions/scaleset"
"github.com/prometheus/client_golang/prometheus"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"sigs.k8s.io/yaml"
)

// The chart values files document the listener metrics as a commented-out
// example. The example lists every metric and every label the exporter can
// populate, with the default buckets; the labels it marks as "not exposed by
// default" are exactly the ones missing from defaultMetrics.
func TestChartValuesMetricsExample(t *testing.T) {
tests := map[string]struct {
path string
key string
}{
"gha-runner-scale-set": {
path: filepath.Join("..", "..", "..", "charts", "gha-runner-scale-set", "values.yaml"),
key: "listenerMetrics",
},
"gha-runner-scale-set-experimental": {
path: filepath.Join("..", "..", "..", "charts", "gha-runner-scale-set-experimental", "values.yaml"),
key: "metrics",
},
}

// Labels documented in the values files as not being part of the defaults.
optInLabels := []string{labelKeyJobWorkflowRef, labelKeyJobWorkflowName, labelKeyJobWorkflowTarget}

// Every label the exporter populates, per metric.
e := &exporter{
scaleSetLabels: prometheus.Labels{
labelKeyRunnerScaleSetName: "",
labelKeyRunnerScaleSetNamespace: "",
labelKeyEnterprise: "",
labelKeyOrganization: "",
labelKeyRepository: "",
},
}
startedLabels := keys(e.startedJobLabels(&scaleset.JobStarted{}))
completedLabels := keys(e.completedJobLabels(&scaleset.JobCompleted{}))
allLabels := map[string][]string{
MetricStartedJobsTotal: startedLabels,
MetricCompletedJobsTotal: completedLabels,
MetricJobStartupDurationSeconds: startedLabels,
MetricJobExecutionDurationSeconds: completedLabels,
}
for name := range defaultMetrics.Gauges {
allLabels[name] = keys(e.scaleSetLabels)
}

for name, tc := range tests {
t.Run(name, func(t *testing.T) {
got := commentedMetricsExample(t, tc.path, tc.key)

require.ElementsMatch(t, keys(defaultMetrics.Counters), keys(got.Counters))
for name, want := range defaultMetrics.Counters {
assertExampleLabels(t, "counter "+name, got.Counters[name].Labels, allLabels[name], want.Labels, optInLabels)
}

require.ElementsMatch(t, keys(defaultMetrics.Gauges), keys(got.Gauges))
for name, want := range defaultMetrics.Gauges {
assertExampleLabels(t, "gauge "+name, got.Gauges[name].Labels, allLabels[name], want.Labels, nil)
}

require.ElementsMatch(t, keys(defaultMetrics.Histograms), keys(got.Histograms))
for name, want := range defaultMetrics.Histograms {
assertExampleLabels(t, "histogram "+name, got.Histograms[name].Labels, allLabels[name], want.Labels, optInLabels)
assert.Equal(t, want.Buckets, got.Histograms[name].Buckets, "histogram %q buckets", name)
}
})
}
}

// assertExampleLabels checks that the example lists every label the exporter
// populates for a metric, and that the labels beyond the defaults are exactly
// the documented opt-in ones.
func assertExampleLabels(t *testing.T, metric string, example, all, defaults, optIn []string) {
t.Helper()

assert.ElementsMatch(t, all, example, "%s labels", metric)
assert.Subset(t, example, defaults, "%s default labels", metric)

var extra []string
for _, l := range example {
if !slices.Contains(defaults, l) {
extra = append(extra, l)
}
}
assert.ElementsMatch(t, optIn, extra, "%s labels not exposed by default", metric)
}

// commentedMetricsExample uncomments the "# <key>:" block of the values file
// and decodes it into a MetricsConfig.
func commentedMetricsExample(t *testing.T, path, key string) v1alpha1.MetricsConfig {
t.Helper()

f, err := os.Open(path)
require.NoError(t, err)
defer f.Close()

var (
lines []string
indent string
)
scanner := bufio.NewScanner(f)
for scanner.Scan() {
line := scanner.Text()
if lines == nil {
if strings.TrimLeft(line, " ") != "# "+key+":" {
continue
}
indent = line[:strings.Index(line, "#")]
}
if !strings.HasPrefix(line, indent+"#") {
break
}
lines = append(lines, strings.TrimPrefix(strings.TrimPrefix(line, indent+"#"), " "))
}
require.NoError(t, scanner.Err())
require.NotEmpty(t, lines, "commented %q block not found in %s", key, path)

var values map[string]v1alpha1.MetricsConfig
require.NoError(t, yaml.UnmarshalStrict([]byte(strings.Join(lines, "\n")), &values))
require.Contains(t, values, key)
return values[key]
}

func keys[V any](m map[string]V) []string {
out := make([]string, 0, len(m))
for k := range m {
out = append(out, k)
}
return out
}