Files
cozystack/internal/controller/workload_controller.go
Timofei Larkin 34991d2cdb Fix virtual machine resource tracking (#904)
* Count Workload resources for pods by requests, not limits
* Do not count init container requests
* Prefix Workloads for pods with `pod-`, just like the other types to
prevent possible name collisions (closes #787)

The previous version of the WorkloadMonitor controller incorrectly
summed resource limits on pods, rather than requests. This prevented it
from tracking the resource allocation for pods, which only had requests
specified, which is particularly the case for kubevirt's virtual machine
pods. Additionally, it counted the limits for all containers, including
init containers, which are short-lived and do not contribute much to the
total resource usage.

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

- **Bug Fixes**
- Improved handling of workloads with unrecognized prefixes by ensuring
they are properly deleted and not processed further.
- Corrected resource aggregation for Pods to sum container resource
requests instead of limits, and now only includes normal containers.

- **New Features**
	- Added support for monitoring workloads with names prefixed by "pod-".

- **Tests**
- Introduced unit tests to verify correct handling of workload name
prefixes and monitored object creation.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

(cherry picked from commit 1e59e5fbb6)
Signed-off-by: Timofei Larkin <lllamnyp@gmail.com>
2025-05-05 13:55:06 +03:00

100 lines
2.5 KiB
Go

package controller
import (
"context"
"strings"
corev1 "k8s.io/api/core/v1"
apierrors "k8s.io/apimachinery/pkg/api/errors"
"k8s.io/apimachinery/pkg/runtime"
"k8s.io/apimachinery/pkg/types"
ctrl "sigs.k8s.io/controller-runtime"
"sigs.k8s.io/controller-runtime/pkg/client"
"sigs.k8s.io/controller-runtime/pkg/log"
cozyv1alpha1 "github.com/cozystack/cozystack/api/v1alpha1"
)
// WorkloadMonitorReconciler reconciles a WorkloadMonitor object
type WorkloadReconciler struct {
client.Client
Scheme *runtime.Scheme
}
func (r *WorkloadReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ctrl.Result, error) {
logger := log.FromContext(ctx)
w := &cozyv1alpha1.Workload{}
err := r.Get(ctx, req.NamespacedName, w)
if err != nil {
if apierrors.IsNotFound(err) {
return ctrl.Result{}, nil
}
logger.Error(err, "Unable to fetch Workload")
return ctrl.Result{}, err
}
// it's being deleted, nothing to handle
if w.DeletionTimestamp != nil {
return ctrl.Result{}, nil
}
t := getMonitoredObject(w)
if t == nil {
err = r.Delete(ctx, w)
if err != nil {
logger.Error(err, "failed to delete workload")
}
return ctrl.Result{}, err
}
err = r.Get(ctx, types.NamespacedName{Name: t.GetName(), Namespace: t.GetNamespace()}, t)
// found object, nothing to do
if err == nil {
return ctrl.Result{}, nil
}
// error getting object but not 404 -- requeue
if !apierrors.IsNotFound(err) {
logger.Error(err, "failed to get dependent object", "kind", t.GetObjectKind(), "dependent-object-name", t.GetName())
return ctrl.Result{}, err
}
err = r.Delete(ctx, w)
if err != nil {
logger.Error(err, "failed to delete workload")
}
return ctrl.Result{}, err
}
// SetupWithManager registers our controller with the Manager and sets up watches.
func (r *WorkloadReconciler) SetupWithManager(mgr ctrl.Manager) error {
return ctrl.NewControllerManagedBy(mgr).
// Watch WorkloadMonitor objects
For(&cozyv1alpha1.Workload{}).
Complete(r)
}
func getMonitoredObject(w *cozyv1alpha1.Workload) client.Object {
switch {
case strings.HasPrefix(w.Name, "pvc-"):
obj := &corev1.PersistentVolumeClaim{}
obj.Name = strings.TrimPrefix(w.Name, "pvc-")
obj.Namespace = w.Namespace
return obj
case strings.HasPrefix(w.Name, "svc-"):
obj := &corev1.Service{}
obj.Name = strings.TrimPrefix(w.Name, "svc-")
obj.Namespace = w.Namespace
return obj
case strings.HasPrefix(w.Name, "pod-"):
obj := &corev1.Pod{}
obj.Name = strings.TrimPrefix(w.Name, "pod-")
obj.Namespace = w.Namespace
return obj
}
var obj client.Object
return obj
}