Merge pull request #15258 from OdedViner/pruner_tollerations

core: add tolerations to crashcollector pruner cronJob pod
This commit is contained in:
Travis Nielsen authored and GitHub committed 2025-01-16 10:00:34 -07:00
commit 53c419afdd
3 files changed
+29 -8

No files matched your search

@@ -32,7 +32,7 @@ import (
"sigs.k8s.io/controller-runtime/pkg/controller/controllerutil"
)
func (r *ReconcileNode) reconcileCrashPruner(namespace string, cephCluster cephv1.CephCluster) error {
func (r *ReconcileNode) reconcileCrashPruner(namespace string, cephCluster cephv1.CephCluster, tolerations []corev1.Toleration) error {
if cephCluster.Spec.CrashCollector.Disable {
logger.Debugf("crash collector is disabled in namespace %q so skipping crash retention reconcile", namespace)
return nil
@@ -60,7 +60,7 @@ func (r *ReconcileNode) reconcileCrashPruner(namespace string, cephCluster cephv
}
} else {
logger.Debugf("daysToRetain set to: %d", cephCluster.Spec.CrashCollector.DaysToRetain)
op, err := r.createOrUpdateCephCron(cephCluster)
op, err := r.createOrUpdateCephCron(cephCluster, tolerations)
if err != nil {
return errors.Wrapf(err, "node reconcile failed on op %q", op)
}
@@ -68,7 +68,7 @@ func (r *ReconcileNode) reconcileCrashPruner(namespace string, cephCluster cephv
}
return nil
}
func (r *ReconcileNode) createOrUpdateCephCron(cephCluster cephv1.CephCluster) (controllerutil.OperationResult, error) {
func (r *ReconcileNode) createOrUpdateCephCron(cephCluster cephv1.CephCluster, tolerations []corev1.Toleration) (controllerutil.OperationResult, error) {
objectMeta := metav1.ObjectMeta{
Name: prunerName,
Namespace: cephCluster.GetNamespace(),
@@ -82,7 +82,6 @@ func (r *ReconcileNode) createOrUpdateCephCron(cephCluster cephv1.CephCluster) (
k8sutil.AppAttr: prunerName,
}
cronJobLabels[k8sutil.ClusterAttr] = cephCluster.GetNamespace()
podTemplateSpec := corev1.PodTemplateSpec{
ObjectMeta: metav1.ObjectMeta{
Labels: cronJobLabels,
@@ -96,6 +95,7 @@ func (r *ReconcileNode) createOrUpdateCephCron(cephCluster cephv1.CephCluster) (
Volumes: volumes,
SecurityContext: &corev1.PodSecurityContext{},
ServiceAccountName: k8sutil.DefaultServiceAccount,
Tolerations: tolerations,
},
}
@@ -27,6 +27,7 @@ import (
"github.com/rook/rook/pkg/operator/test"
"github.com/stretchr/testify/assert"
v1 "k8s.io/api/batch/v1"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/types"
"sigs.k8s.io/controller-runtime/pkg/client/fake"
@@ -34,7 +35,26 @@ import (
)
func TestCreateOrUpdateCephCron(t *testing.T) {
cephCluster := cephv1.CephCluster{ObjectMeta: metav1.ObjectMeta{Namespace: "rook-ceph"}}
tolerations := []corev1.Toleration{
{
Key: "key",
Operator: corev1.TolerationOpEqual,
Value: "value",
Effect: corev1.TaintEffectNoSchedule,
},
}
cephCluster := cephv1.CephCluster{
ObjectMeta: metav1.ObjectMeta{
Namespace: "rook-ceph",
},
Spec: cephv1.ClusterSpec{
Placement: cephv1.PlacementSpec{
"all": {
Tolerations: tolerations,
},
},
},
}
ctx := context.TODO()
context := &clusterd.Context{
Clientset: test.New(t, 1),
@@ -61,10 +81,11 @@ func TestCreateOrUpdateCephCron(t *testing.T) {
}
// check if cronJob is created
cntrlutil, err := r.createOrUpdateCephCron(cephCluster)
cntrlutil, err := r.createOrUpdateCephCron(cephCluster, tolerations)
assert.NoError(t, err)
assert.Equal(t, cntrlutil, controllerutil.OperationResult("created"))
err = r.client.Get(ctx, types.NamespacedName{Namespace: "rook-ceph", Name: prunerName}, cronV1)
assert.NoError(t, err)
assert.Equal(t, tolerations, cronV1.Spec.JobTemplate.Spec.Template.Spec.Tolerations, "Tolerations do not match")
}
@@ -190,8 +190,8 @@ func (r *ReconcileNode) reconcile(request reconcile.Request) (reconcile.Result,
}
// If the node has Ceph pods we create the daemons
tolerations := uniqueTolerations.ToList()
if hasCephPods {
tolerations := uniqueTolerations.ToList()
err := r.createOrUpdateNodeDaemons(*node, tolerations, cephCluster, cephVersion)
if err != nil {
return reconcile.Result{}, errors.Wrap(err, "node reconcile failed")
@@ -219,7 +219,7 @@ func (r *ReconcileNode) reconcile(request reconcile.Request) (reconcile.Result,
}
}
if err := r.reconcileCrashPruner(namespace, cephCluster); err != nil {
if err := r.reconcileCrashPruner(namespace, cephCluster, tolerations); err != nil {
return reconcile.Result{}, err
}
}