System Test, RDS Core: metalLB traffic segregation test suite was added

[elenagerman]

No description provided.

[elenagerman]

@yprokule PTAL

[yprokule]

I'd rather expose replica count as a parameter to cover a use case where a pod(s) can ran on multiple nodes but different amount of replicas is required (for e.g. a pod could run on node-x, node-y, node-z, but app requires only 2 instances).
Anyway that's only a suggestion so leaving it up to U

[elenagerman]

I'm sorry, but I'm not sure it's a good idea. different nodes/pods count is not exactly our case.
Also, the question is, why provide more nodes than you need pods?
In our case, for both tcpdump and traceroute scenarios, we provide precisely the nodes list where we need to deploy the support tools pods; I mean, we need to monitor a specific node => we need a pod on this node.
I would leave it like this right now if you don't mind.
If we have any future use cases with different requirements, we can modify it. But without a specific proposal for different nodes and pod counts, it's hard to understand how it would be better to change it.

[yprokule]

where's the code to get the node name? assume smth like:

lbOneNodeName = lbOnePodsList[0].Object.Spec.NodeName

[yprokule]

Suggested change

	glog.V(rdscoreparams.RDSCoreLogLevel).Infof("Searching for pods in %q namespace",
	RDSCoreConfig.MetalLBLoadBalancerTwoNamespace)
	glog.V(rdscoreparams.RDSCoreLogLevel).Infof("Searching for pods in %q namespace with %q label",
	RDSCoreConfig.MetalLBLoadBalancerTwoNamespace, lbDeploymentLabel)

[yprokule]

ditto, what happens if 2 pods found? is it safe to ignore all other pods and always use 1st one?

[yprokule]

the command is show ip(v6) route bgp so not sure why it's stated that route for requested IP address was present in the output of the aforementioned command:

Suggested change

	glog.V(100).Infof(fmt.Sprintf("the IP routes %s were found on the FRR container %s: %s",
	lbIP, containerName, result))
	glog.V(100).Infof(fmt.Sprintf("BGP routes were found in the FRR container %s\n: %s",
	containerName, result))

[yprokule]

Suggested change

	glog.V(100).Infof("Failed to verify IP route bgp on the FRR container %s: %v", containerName, err)
	glog.V(100).Infof("Failed to verify BGP routes in the FRR container %s: %v", containerName, err)

[yprokule]

Suggested change

	return fmt.Errorf("failed to verify IP route bgp on the FRR container %s: %w", containerName, err)
	return fmt.Errorf("failed to verify BGP routes in the FRR container %s: %w", containerName, err)

[yprokule]

Suggested change

	glog.V(100).Infof("No IP route bgp %s on the FRR container %s was found", lbIP, containerName)
	glog.V(100).Infof("No BGP route %s in the FRR container %s was found", lbIP, containerName)

[yprokule]

Suggested change

	return fmt.Errorf("no IP route bgp %s on the FRR container %s was found", lbIP, containerName)
	return fmt.Errorf("no BGP route %s in the FRR container %s was found", lbIP, containerName)

[yprokule]

the function is already retried so why this extra for loop?