Skip to content

Metric conditions evaluation hotfix and check_drivesize tests #428

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

Closed
inqrphl wants to merge 9 commits into from
Closed

Metric conditions evaluation hotfix and check_drivesize tests #428

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
ba35024
metric: fix CheckMetric.CheckForThresholds
Jul 23, 2026
8a38ba0
checks: CheckData.CheckMetrics uses the metric's condition list, whic…
Jul 23, 2026
7f7d6ee
check_drivesize: support keywords using "used %" , "free %" "inodes_u…
Jul 23, 2026
5b358e5
Merge branch 'main' into metric-conditions-evaluation-hotfix
inqrphl Jul 23, 2026
5078ecc
check_temperature: expand the temp and temperature metric macros prop…
Jul 23, 2026
0b4da0f
check_drivesize: add tests with two drives and specialized drive cond…
Jul 23, 2026
6df7499
check_drivesize: skip specialized condition perfdata check on freebsd…
Jul 23, 2026
f277a3b
check_drivesize: completely disable specialized drive conditions perf…
Jul 23, 2026
353587b
checkdata: remove unused warning and critical thresholds arguments in…
Jul 23, 2026
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
25 changes: 13 additions & 12 deletions pkg/snclient/check_drivesize.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -356,36 +356,37 @@ func (l *CheckDrivesize) addMetrics(metricPrefix string, drive map[string]string
}

driveFreeMetricLabel := fmt.Sprintf("%s free %%", metricPrefix)
if check.HasThreshold(driveFreeMetricLabel) || check.HasThreshold("free") || check.HasThreshold("free_pct") || check.HasThreshold("free_bytes") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"free_pct", "free_bytes"}, "free", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"free_pct", "free_bytes"}, "free", check.critThreshold)
if check.HasThreshold(driveFreeMetricLabel) || check.HasThreshold("free") || check.HasThreshold("free_pct") || check.HasThreshold("free_bytes") || check.HasThreshold("free %") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"free_pct", "free_bytes", "free %"}, "free", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"free_pct", "free_bytes", "free %"}, "free", check.critThreshold)
perfLabel := fmt.Sprintf("%s free", metricPrefix)
check.AddBytePercentMetrics("free", perfLabel, magic*float64(usage.Free), magic*float64(total), drive)
check.processMetricsWithSpecializedKeyword("drive", perfLabel, drive)
}

driveUsagePctMetricLabel := fmt.Sprintf("%s used %%", metricPrefix)
if check.HasThreshold(driveUsagePctMetricLabel) || check.HasThreshold("used") || check.HasThreshold("used_pct") || check.HasThreshold("used_bytes") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"used_pct", "used_bytes"}, "used", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"used_pct", "used_bytes"}, "used", check.critThreshold)
if check.HasThreshold(driveUsagePctMetricLabel) || check.HasThreshold("used") || check.HasThreshold("used_pct") || check.HasThreshold("used_bytes") || check.HasThreshold("used %") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"used_pct", "used_bytes", "used %"}, "used", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"used_pct", "used_bytes", "used %"}, "used", check.critThreshold)
perfLabel := fmt.Sprintf("%s used", metricPrefix)
check.AddBytePercentMetrics("used", perfLabel, magic*float64(usage.Used), magic*float64(total), drive)
check.processMetricsWithSpecializedKeyword("drive", perfLabel, drive)
}

driveInodesUsedPctMetricLabel := fmt.Sprintf("%s inodes_used %%", metricPrefix)
if check.HasThreshold(driveInodesUsedPctMetricLabel) || check.HasThreshold("inodes") || check.HasThreshold("inodes_used") || check.HasThreshold("inodes_used_pct") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"inodes_used_pct"}, "inodes_used", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"inodes_used_pct"}, "inodes_used", check.critThreshold)
if check.HasThreshold(driveInodesUsedPctMetricLabel) || check.HasThreshold("inodes") || check.HasThreshold("inodes_used") || check.HasThreshold("inodes_used_pct") ||
check.HasThreshold("inodes_used %") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"inodes_used_pct", "inodes_used %"}, "inodes_used", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"inodes_used_pct", "inodes_used %"}, "inodes_used", check.critThreshold)
perfLabel := fmt.Sprintf("%s inodes_used", metricPrefix)
check.AddPercentMetrics("inodes_used", perfLabel, float64(usage.InodesUsed), float64(usage.InodesTotal), drive)
check.processMetricsWithSpecializedKeyword("drive", perfLabel, drive)
}

driveInodesFreePctMetricLabel := fmt.Sprintf("%s inodes_free %%", metricPrefix)
if check.HasThreshold(driveInodesFreePctMetricLabel) || check.HasThreshold("inodes_free") || check.HasThreshold("inodes_free_pct") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"inodes_free_pct"}, "inodes_free", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"inodes_free_pct"}, "inodes_free", check.critThreshold)
if check.HasThreshold(driveInodesFreePctMetricLabel) || check.HasThreshold("inodes_free") || check.HasThreshold("inodes_free_pct") || check.HasThreshold("inodes_free %") {
check.warnThreshold = check.TransformMultipleKeywords([]string{"inodes_free_pct", "inodes_free %"}, "inodes_free", check.warnThreshold)
check.critThreshold = check.TransformMultipleKeywords([]string{"inodes_free_pct", "inodes_free %"}, "inodes_free", check.critThreshold)
perfLabel := fmt.Sprintf("%s inodes_free", metricPrefix)
check.AddPercentMetrics("inodes_free", perfLabel, float64(usage.InodesFree), float64(usage.InodesTotal), drive)
check.processMetricsWithSpecializedKeyword("drive", perfLabel, drive)
Expand Down
66 changes: 66 additions & 0 deletions pkg/snclient/check_drivesize_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,3 +83,69 @@ func TestNonexistingDrive(t *testing.T) {

StopTestAgent(t, snc)
}

func TestRootDriveThresholdsDrivePercentSign(t *testing.T) {
snc := StartTestAgent(t, "")

// This tests the conditions using '[drive] used %' keyword
res := snc.RunCheck("check_drivesize", []string{"drive=/", "warn='/ used %' >= 0", "crit=none"})
assert.Equalf(t, CheckExitWarning, res.State, "state should be WARNING")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "crit='/ used %' >= 0"})
assert.Equalf(t, CheckExitCritical, res.State, "state should be CRITICAL")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "warn='/ used %' >= 0", "crit='/ used %' >= 0"})
assert.Equalf(t, CheckExitCritical, res.State, "state should be CRITICAL")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "warn='/ used %' >= 0", "crit='/ used %' >= 0", "ok='/ used %' >= 0"})
assert.Equalf(t, CheckExitOK, res.State, "state should be CRITICAL")

StopTestAgent(t, snc)
}

func TestRootDriveThresholdsUsedPct(t *testing.T) {
snc := StartTestAgent(t, "")

// This tests the conditions using '[drive] used_pct' keyword
res := snc.RunCheck("check_drivesize", []string{"drive=/", "warn='/ used_pct' >= 0", "crit=none"})
assert.Equalf(t, CheckExitWarning, res.State, "state should be WARNING")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "crit='/ used_pct' >= 0"})
assert.Equalf(t, CheckExitCritical, res.State, "state should be CRITICAL")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "warn='/ used_pct' >= 0", "crit='/ used_pct' >= 0"})
assert.Equalf(t, CheckExitCritical, res.State, "state should be CRITICAL")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "warn='/ used_pct' >= 0", "crit='/ used_pct' >= 0", "ok='/ used_pct' >= 0"})
assert.Equalf(t, CheckExitOK, res.State, "state should be CRITICAL")

StopTestAgent(t, snc)
}

func TestRootDriveThresholdsUsedPercentSign(t *testing.T) {
snc := StartTestAgent(t, "")

// Generic 'used %' keyword, should trigger
res := snc.RunCheck("check_drivesize", []string{"drive=/", "warning='used %' gt 0", "critical=none"})
assert.Equalf(t, CheckExitWarning, res.State, "state should be WARNING")

res = snc.RunCheck("check_drivesize", []string{"drive=/", "critical='used %' gt 0"})
assert.Equalf(t, CheckExitCritical, res.State, "state should be CRITICAL")

StopTestAgent(t, snc)
}

func TestRootDriveSpecializedDriveConditions(t *testing.T) {
snc := StartTestAgent(t, "")

// In critical conditions, the second one is filtered as it contains the drive="/" keyword.
// Since such a condition is present, other critical conditions are filtered out for that entry
res := snc.RunCheck("check_drivesize", []string{"drive=/", "critical='used %' gt 0", "critical=drive eq '/' and 'used_pct' gt 100"})
assert.Equalf(t, CheckExitOK, res.State, "state should be OK")

// Filtering does not work here, since the specialized condition is in the warning threshold, not critical threshold
res = snc.RunCheck("check_drivesize", []string{"drive=/", "critical='used %' gt 0", "warning=drive eq '/' and 'used_pct' gt 0"})
assert.Equalf(t, CheckExitCritical, res.State, "state should be WARNING")

StopTestAgent(t, snc)
}
9 changes: 7 additions & 2 deletions pkg/snclient/check_temperature.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,14 +124,19 @@ func (l *CheckTemperature) addSensor(check *CheckData, sensor *temperatureStat,

minVal := utils.ToPrecision(sensor.Min, 3)
maxVal := utils.ToPrecision(sensor.High, 3)
var warning, critical ConditionList
if sensor.Min != 0 || sensor.Critical != 0 {

@sni sni Jul 23, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

how is the expansion of temp and temperature related to Min/Critical? I don't understand why this is necessary.

@inqrphl inqrphl Jul 23, 2026

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.

The expansion has nothing to do with it, I am only adding the modified warning/critical thresholds if sensor.Min != 0 or sensor.Critical != 0

Some sensors report min==0 and then get added, only to fail directly in the check_temperature_test

Of course that problem only started happening after the CheckMetric.CheckForThresholds was fixed in the pr.

warning = check.ExpandMetricMacros(check.TransformMultipleKeywords([]string{"temp", "temperature"}, sensorKey, check.warnThreshold), entry)
critical = check.ExpandMetricMacros(check.TransformMultipleKeywords([]string{"temp", "temperature"}, sensorKey, check.critThreshold), entry)
}
check.result.Metrics = append(check.result.Metrics, &CheckMetric{
ThresholdName: sensorKey,
Name: sensorKey,
Value: utils.ToPrecision(sensor.Temperature, 3),
Min: &minVal,
Max: &maxVal,
Warning: check.ExpandMetricMacros(check.TransformMultipleKeywords([]string{"temp", "temperature"}, sensorKey, check.warnThreshold), entry),
Critical: check.ExpandMetricMacros(check.TransformMultipleKeywords([]string{"temp", "temperature"}, sensorKey, check.critThreshold), entry),
Warning: warning,
Critical: critical,
})

check.listData = append(check.listData, entry)
Expand Down
9 changes: 5 additions & 4 deletions pkg/snclient/checkdata.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -239,7 +239,8 @@ func (cd *CheckData) finalizeOutput() (*CheckResult, error) {
cd.setStateFromMaps(finalMacros)
// Metrics are checked last, which also sets the final state
log.Tracef("checking warning, critical, and ok thresholds on check metrics")
cd.CheckMetrics(cd.warnThreshold, cd.critThreshold, cd.okThreshold)
// metrics save their own warning and critical thresholds, but ok thresholds come from check itself
cd.CheckMetrics(cd.okThreshold)

switch {
case cd.result.Output != "":
Expand Down Expand Up @@ -483,16 +484,16 @@ func (cd *CheckData) Check(data map[string]string, warnCond, critCond, okCond Co
}

// CheckMetrics tries warn/crit/ok conditions against given metrics and sets final state accordingly
func (cd *CheckData) CheckMetrics(warnCond, critCond, okCond ConditionList) {
func (cd *CheckData) CheckMetrics(okCond ConditionList) {
// each metric is ran through conditions individually
for _, metric := range cd.result.Metrics {
state := CheckExitOK

if metric.CheckForThresholds(&warnCond) {
if metric.CheckForThresholds(&metric.Warning) {
state = CheckExitWarning
}

if metric.CheckForThresholds(&critCond) {
if metric.CheckForThresholds(&metric.Critical) {
state = CheckExitCritical
}

Expand Down
4 changes: 2 additions & 2 deletions pkg/snclient/checkmetric.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -191,13 +191,13 @@ func (m *CheckMetric) BuildCheckData() (data map[string]string) {
return data
}

// Metrics have a defined way for specifying key / values
// Returns if at least one condition in the conditionList is found to be correct
//
//nolint:dogsled // only need corrects list here
func (m *CheckMetric) CheckForThresholds(conditionList *ConditionList) (result bool) {
data := m.BuildCheckData()

corrects, _, _, _ := conditionList.performMatches(data, true)

return len(corrects) > 1
return len(corrects) > 0
}