Kubernetes SD: disable resync timer

While originally the resync period also forced refreshing from Kubernetes API server, this has been removed for some years now because watching the API server got more stable [1]. Today, this just results in all entities being sent to the service discovery again, which is valid from a general Prometheus perspective, but results in unnecessary CPU load and also breaks service discovery metrics. In especially, this makes monitoring "do we actually observe changes from Kubernetes API server" impossible (receiving constant updates from Kubernetes service discovery is a pretty valid assumption, for example nodes get frequent status updates, ...).

Signed-off-by: Jens Erat <jens.erat@mercedes-benz.com>
This commit is contained in:
Jens Erat 2022-12-20 12:40:04 +01:00
parent b2226258bc
commit 728fdc959e
No known key found for this signature in database
GPG Key ID: AA8F3A2591A041DB

View File

@ -382,7 +382,8 @@ func mapSelector(rawSelector []SelectorConfig) roleSelector {
return rs
}
const resyncPeriod = 10 * time.Minute
// Disable the informer's resync, which just periodically resends already processed updates and distort SD metrics.
const resyncDisabled = 0
// Run implements the discoverer interface.
func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) {
@ -475,8 +476,8 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) {
eps := NewEndpointSlice(
log.With(d.logger, "role", "endpointslice"),
informer,
cache.NewSharedInformer(slw, &apiv1.Service{}, resyncPeriod),
cache.NewSharedInformer(plw, &apiv1.Pod{}, resyncPeriod),
cache.NewSharedInformer(slw, &apiv1.Service{}, resyncDisabled),
cache.NewSharedInformer(plw, &apiv1.Pod{}, resyncDisabled),
nodeInf,
)
d.discoverers = append(d.discoverers, eps)
@ -534,8 +535,8 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) {
eps := NewEndpoints(
log.With(d.logger, "role", "endpoint"),
d.newEndpointsByNodeInformer(elw),
cache.NewSharedInformer(slw, &apiv1.Service{}, resyncPeriod),
cache.NewSharedInformer(plw, &apiv1.Pod{}, resyncPeriod),
cache.NewSharedInformer(slw, &apiv1.Service{}, resyncDisabled),
cache.NewSharedInformer(plw, &apiv1.Pod{}, resyncDisabled),
nodeInf,
)
d.discoverers = append(d.discoverers, eps)
@ -589,7 +590,7 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) {
}
svc := NewService(
log.With(d.logger, "role", "service"),
cache.NewSharedInformer(slw, &apiv1.Service{}, resyncPeriod),
cache.NewSharedInformer(slw, &apiv1.Service{}, resyncDisabled),
)
d.discoverers = append(d.discoverers, svc)
go svc.informer.Run(ctx.Done())
@ -627,7 +628,7 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) {
return i.Watch(ctx, options)
},
}
informer = cache.NewSharedInformer(ilw, &networkv1.Ingress{}, resyncPeriod)
informer = cache.NewSharedInformer(ilw, &networkv1.Ingress{}, resyncDisabled)
} else {
i := d.client.NetworkingV1beta1().Ingresses(namespace)
ilw := &cache.ListWatch{
@ -642,7 +643,7 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) {
return i.Watch(ctx, options)
},
}
informer = cache.NewSharedInformer(ilw, &v1beta1.Ingress{}, resyncPeriod)
informer = cache.NewSharedInformer(ilw, &v1beta1.Ingress{}, resyncDisabled)
}
ingress := NewIngress(
log.With(d.logger, "role", "ingress"),
@ -732,7 +733,7 @@ func (d *Discovery) newNodeInformer(ctx context.Context) cache.SharedInformer {
return d.client.CoreV1().Nodes().Watch(ctx, options)
},
}
return cache.NewSharedInformer(nlw, &apiv1.Node{}, resyncPeriod)
return cache.NewSharedInformer(nlw, &apiv1.Node{}, resyncDisabled)
}
func (d *Discovery) newPodsByNodeInformer(plw *cache.ListWatch) cache.SharedIndexInformer {
@ -747,13 +748,13 @@ func (d *Discovery) newPodsByNodeInformer(plw *cache.ListWatch) cache.SharedInde
}
}
return cache.NewSharedIndexInformer(plw, &apiv1.Pod{}, resyncPeriod, indexers)
return cache.NewSharedIndexInformer(plw, &apiv1.Pod{}, resyncDisabled, indexers)
}
func (d *Discovery) newEndpointsByNodeInformer(plw *cache.ListWatch) cache.SharedIndexInformer {
indexers := make(map[string]cache.IndexFunc)
if !d.attachMetadata.Node {
return cache.NewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncPeriod, indexers)
return cache.NewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncDisabled, indexers)
}
indexers[nodeIndex] = func(obj interface{}) ([]string, error) {
@ -773,13 +774,13 @@ func (d *Discovery) newEndpointsByNodeInformer(plw *cache.ListWatch) cache.Share
return nodes, nil
}
return cache.NewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncPeriod, indexers)
return cache.NewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncDisabled, indexers)
}
func (d *Discovery) newEndpointSlicesByNodeInformer(plw *cache.ListWatch, object runtime.Object) cache.SharedIndexInformer {
indexers := make(map[string]cache.IndexFunc)
if !d.attachMetadata.Node {
cache.NewSharedIndexInformer(plw, &disv1.EndpointSlice{}, resyncPeriod, indexers)
cache.NewSharedIndexInformer(plw, &disv1.EndpointSlice{}, resyncDisabled, indexers)
}
indexers[nodeIndex] = func(obj interface{}) ([]string, error) {
@ -806,7 +807,7 @@ func (d *Discovery) newEndpointSlicesByNodeInformer(plw *cache.ListWatch, object
return nodes, nil
}
return cache.NewSharedIndexInformer(plw, object, resyncPeriod, indexers)
return cache.NewSharedIndexInformer(plw, object, resyncDisabled, indexers)
}
func checkDiscoveryV1Supported(client kubernetes.Interface) (bool, error) {