Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add LatencyNative availability metric #1417

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Add LatencyNative availability metric #1417

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8cf169b
Add LatencyNative availability
samwright Feb 24, 2025
56a3896
add tests
samwright Feb 24, 2025
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
47 changes: 44 additions & 3 deletions slo/rules.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -512,7 +512,7 @@ func (o Objective) Burnrate(timerange time.Duration) string {

return expr.String()
case LatencyNative:
expr, err := parser.ParseExpr(`1 - histogram_fraction(0,0.696969, rate(metric{matchers="total"}[1s]))`)
expr, err := parser.ParseExpr(`1 - histogram_fraction(0,0.696969, sum(rate(metric{matchers="total"}[1s])))`)
if err != nil {
return err.Error()
}
Expand Down Expand Up @@ -911,7 +911,7 @@ func (o Objective) IncreaseRules() (monitoringv1.RuleGroup, error) {
}
}

expr, err := parser.ParseExpr(`histogram_count(increase(metric{matchers="total"}[1s]))`)
expr, err := parser.ParseExpr(`histogram_count(sum(increase(metric{matchers="total"}[1s])))`)
if err != nil {
return monitoringv1.RuleGroup{}, err
}
Expand All @@ -929,7 +929,7 @@ func (o Objective) IncreaseRules() (monitoringv1.RuleGroup, error) {
Labels: ruleLabels,
})

expr, err = parser.ParseExpr(`histogram_fraction(0, 0.696969, increase(metric{matchers="total"}[1s])) * histogram_count(increase(metric{matchers="total"}[1s]))`)
expr, err = parser.ParseExpr(`histogram_fraction(0, 0.696969, sum(increase(metric{matchers="total"}[1s]))) * histogram_count(sum(increase(metric{matchers="total"}[1s])))`)
if err != nil {
return monitoringv1.RuleGroup{}, err
}
Expand Down Expand Up @@ -1384,6 +1384,47 @@ func (o Objective) GenericRules() (monitoringv1.RuleGroup, error) {
Labels: ruleLabels,
})
}
case LatencyNative:
latencySeconds := time.Duration(o.Indicator.LatencyNative.Latency).Seconds()

// availability
{
expr, err := parser.ParseExpr(`sum(metric{matchers="errors"} or vector(0)) / sum(metric{matchers="total"})`)
if err != nil {
return monitoringv1.RuleGroup{}, err
}

metric := increaseName(o.Indicator.LatencyNative.Total.Name, o.Window)
matchers := o.Indicator.LatencyNative.Total.LabelMatchers
for _, m := range matchers {
if m.Name == labels.MetricName {
m.Value = metric
break
}
}
matchers = append(matchers, &labels.Matcher{
Type: labels.MatchEqual,
Name: "slo",
Value: o.Name(),
})

errorMatchers := slices.Clone(matchers)
errorMatchers = append(errorMatchers, &labels.Matcher{Type: labels.MatchEqual, Name: "le", Value: fmt.Sprintf("%g", latencySeconds)})
matchers = append(matchers, &labels.Matcher{Type: labels.MatchEqual, Name: "le", Value: ""})

objectiveReplacer{
metric: metric,
matchers: matchers,
errorMatchers: errorMatchers,
window: time.Duration(o.Window),
}.replace(expr)

rules = append(rules, monitoringv1.Rule{
Record: "pyrra_availability",
Expr: intstr.FromString(expr.String()),
Labels: ruleLabels,
})
}

case BoolGauge:
if len(o.Indicator.BoolGauge.Grouping) > 0 {
Expand Down
40 changes: 30 additions & 10 deletions slo/rules_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -359,31 +359,31 @@ func TestObjective_Burnrates(t *testing.T) {
Interval: monitoringDuration("30s"),
Rules: []monitoringv1.Rule{{
Record: "http_request_duration_seconds:burnrate5m",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[5m]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[5m])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:burnrate30m",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[30m]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[30m])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:burnrate1h",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[1h]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[1h])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:burnrate2h",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[2h]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[2h])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:burnrate6h",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[6h]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[6h])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:burnrate1d",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[1d]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[1d])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:burnrate4d",
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4d]))`),
Expr: intstr.FromString(`1 - histogram_fraction(0, 1, sum(rate(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4d])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Alert: "ErrorBudgetBurn",
Expand Down Expand Up @@ -1369,11 +1369,11 @@ func TestObjective_IncreaseRules(t *testing.T) {
Interval: monitoringDuration("2m30s"),
Rules: []monitoringv1.Rule{{
Record: "http_request_duration_seconds:increase4w",
Expr: intstr.FromString(`histogram_count(increase(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4w]))`),
Expr: intstr.FromString(`histogram_count(sum(increase(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4w])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency"},
}, {
Record: "http_request_duration_seconds:increase4w",
Expr: intstr.FromString(`histogram_fraction(0, 1, increase(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4w])) * histogram_count(increase(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4w]))`),
Expr: intstr.FromString(`histogram_fraction(0, 1, sum(increase(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4w]))) * histogram_count(sum(increase(http_request_duration_seconds{code=~"2..",job="metrics-service-thanos-receive-default"}[4w])))`),
Labels: map[string]string{"job": "metrics-service-thanos-receive-default", "slo": "monitoring-http-latency", "le": "1"},
//}, {
// Alert: "SLOMetricAbsent",
Expand Down Expand Up @@ -1774,6 +1774,26 @@ func TestObjective_GrafanaRules(t *testing.T) {
Labels: map[string]string{"slo": "monitoring-http-latency"},
}},
},
}, {
name: "http-latency-native",
slo: objectiveHTTPNativeLatency(),
rules: monitoringv1.RuleGroup{
Name: "monitoring-http-latency-generic",
Interval: monitoringDuration("30s"),
Rules: []monitoringv1.Rule{{
Record: "pyrra_objective",
Expr: intstr.FromString("0.995"),
Labels: map[string]string{"slo": "monitoring-http-latency"},
}, {
Record: "pyrra_window",
Expr: intstr.FromString("2419200"),
Labels: map[string]string{"slo": "monitoring-http-latency"},
}, {
Record: "pyrra_availability",
Expr: intstr.FromString(`sum(http_request_duration_seconds:increase4w{code=~"2..",job="metrics-service-thanos-receive-default",le="1",slo="monitoring-http-latency"} or vector(0)) / sum(http_request_duration_seconds:increase4w{code=~"2..",job="metrics-service-thanos-receive-default",le="",slo="monitoring-http-latency"})`),
Labels: map[string]string{"slo": "monitoring-http-latency"},
}},
},
}, {
name: "http-latency-grouping",
slo: objectiveHTTPLatencyGrouping(),
Expand Down Expand Up @@ -1912,7 +1932,7 @@ func TestObjective_GrafanaRules(t *testing.T) {
err: ErrGroupingUnsupported,
}}

require.Len(t, testcases, 16)
require.Len(t, testcases, 17)

for _, tc := range testcases {
t.Run(tc.name, func(t *testing.T) {
Expand Down
Loading