Skip to content

OCPBUGS-24003: mcd_config_drift not working properly #5632

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

Open
dkhater-redhat wants to merge 1 commit into
base: main
Choose a base branch
from
+39 −1
Open

OCPBUGS-24003: mcd_config_drift not working properly #5632

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
40 changes: 39 additions & 1 deletion pkg/daemon/daemon.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -246,6 +246,10 @@ const (
// Where nmstate writes the link files if it persisted ifnames.
// https://github.com/nmstate/nmstate/blob/03c7b03bd4c9b0067d3811dbbf72635201519356/rust/src/cli/persist_nic.rs#L32-L36
systemdNetworkDir = "etc/systemd/network"

// Config drift error message fragments used to identify drift-related degradation
configDriftContentMismatch = "content mismatch"
configDriftModeMismatch = "mode mismatch"
)

type onceFromOrigin int
Expand Down Expand Up @@ -1522,8 +1526,37 @@ func (dn *Daemon) getCurrentConfigFromNode() (*onDiskConfig, error) {
return tempConfig, nil
}

func (dn *Daemon) startConfigDriftMonitor() {
// initConfigDriftMetric initializes the config drift metric based on the node's current state.
// If the node is Degraded due to config drift, set the metric to current time to indicate ongoing drift.
// Otherwise, clear the metric.
func (dn *Daemon) initConfigDriftMetric() {
var reason string

state, err := getNodeAnnotationExt(dn.node, constants.MachineConfigDaemonStateAnnotationKey, true)
if err != nil {
klog.Warningf("Could not get node state when initializing config drift metric: %v", err)
goto clearMetric
}
if state != constants.MachineConfigDaemonStateDegraded {
goto clearMetric
}

reason, err = getNodeAnnotationExt(dn.node, constants.MachineConfigDaemonReasonAnnotationKey, true)
if err != nil {
klog.Warningf("Could not get node reason when initializing config drift metric: %v", err)
goto clearMetric
}
if strings.Contains(reason, configDriftContentMismatch) || strings.Contains(reason, configDriftModeMismatch) {
mcdConfigDrift.SetToCurrentTime()
klog.Infof("Config drift metric initialized: node is degraded due to config drift")
return
}

clearMetric:
mcdConfigDrift.Set(0)
}

func (dn *Daemon) startConfigDriftMonitor() {
// Even though the Config Drift Monitor object ensures that only a single
// Config Drift Watcher is running at any given time, other things, such as
// emitting Kube events on startup, should only occur if we weren't
Expand All @@ -1533,6 +1566,8 @@ func (dn *Daemon) startConfigDriftMonitor() {
return
}

dn.initConfigDriftMetric()

odc, err := dn.getCurrentConfigOnDisk()
if err != nil && !os.IsNotExist(err) {
dn.exitCh <- fmt.Errorf("could not get current config from disk: %w", err)
Expand Down Expand Up @@ -2291,6 +2326,9 @@ func (dn *Daemon) checkStateOnFirstRun() error {

if err := dn.validateOnDiskStateOrImage(state.currentConfig, state.currentImage); err != nil {
dn.nodeWriter.Eventf(corev1.EventTypeWarning, "OnDiskStateValidationFailed", err.Error())
// Start the config drift monitor even when there's pre-existing drift
// so the metric gets initialized correctly on MCD restart
dn.startConfigDriftMonitor()
return err
}

Expand Down