perf(source): benchmarks on EndpointTargetsFromServices

[ivankatliarchuk]

What does it do ?

In this PR, I've added a benchmark tests, to showcase bottlenecks described here #5473 (comment)

• benchmark tests
• added more tests
• found duplicate code and replaced with function

Motivation

When decided to have a look started

• add workerCount config and multiple worker supports for istio_gateway and istio_virtualservice #5473
• Slow updates with 3K+ objects #5016

And this comment #5493 (comment)

One of the bottlenecks is in this lines

external-dns/source/endpoints.go

Lines 87 to 110 in 051c900

	services, err := svcInformer.Lister().Services(namespace).List(labels.Everything())
	if err != nil {
	return nil, fmt.Errorf("failed to list labels for services in namespace %q: %w", namespace, err)
	}
	for _, service := range services {
	if !MatchesServiceSelector(selector, service.Spec.Selector) {
	continue
	}
	if len(service.Spec.ExternalIPs) > 0 {
	targets = append(targets, service.Spec.ExternalIPs...)
	continue
	}
	for _, lb := range service.Status.LoadBalancer.Ingress {
	if lb.IP != "" {
	targets = append(targets, lb.IP)
	} else if lb.Hostname != "" {
	targets = append(targets, lb.Hostname)
	}
	}
	}
	return targets, nil

Basically it does

• retrive all services from the cache
• iterate over each service
• iterate over each label

On top of that, the function EndpointTargetsFromServices actually called from within the loop of all virtual services or any other kubernetes entity like here. This is quite inneficiate....

Benchmark results

BenchmarkEndpointTargetsFromServicesMedium-16                       	   11194	    105419 ns/op
BenchmarkEndpointTargetsFromServicesMediumIterateOverGateways
BenchmarkEndpointTargetsFromServicesMediumIterateOverGateways-16    	     290	   4070484 ns/op
BenchmarkEndpointTargetsFromServicesHigh
BenchmarkEndpointTargetsFromServicesHigh-16                         	      88	  12669344 ns/op
BenchmarkEndpointTargetsFromServicesHighIterateOverGateways
BenchmarkEndpointTargetsFromServicesHighIterateOverGateways-16      	       1	13425263466 ns/op

Just a hightlight from one of the issues

We are running on EKS with BatchSizeConfig of 1,000 (default), we have ~3K Isito virtualService objects & 1K Service objects.

So this means that there are 3000 * 1000 * (number of selectors), every time there is a reconcile.

Follo-up:

• potential improvement from this PR Speed up things with indexers gofogo/k8s-sigs-external-dns-fork#31

More

• Yes, this PR title follows Conventional Commits
• Yes, I added unit tests
• Yes, I updated end user documentation accordingly

[k8s-ci-robot]

@AndrewCharlesHay: changing LGTM is restricted to collaborators

Details

In response to this:

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[k8s-ci-robot]

@AndrewCharlesHay: changing LGTM is restricted to collaborators

Details

In response to this:

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[ivankatliarchuk]

cc @mloiseleur

[mloiseleur]

/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: AndrewCharlesHay, mloiseleur

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [mloiseleur]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[mloiseleur]

/lgtm