Check if machineDeployment queue exists before using it

[alvaroaleman]

Resolves #501

What this PR does / why we need it:

Avoids NPEs when starting the machineDeployment controller in a cluster that has pre-existing machineSets

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):
Fixes #501

Special notes for your reviewer:

Please confirm that if this PR changes any image versions, then that's the sole change this PR makes.

Release note:

NONE

[alvaroaleman]

/assign @roberthbailey

[xmudrii]

/ok-to-test

[Lion-Wei]

Shouldn't here have an error out after multiple attempts or timeout?

[alvaroaleman]

The only case where this happens is during startup of the controller because init is called before the workqeue is created:

cluster-api/pkg/controller/machinedeployment/zz_generated.api.register.go

Lines 59 to 71 in 5b128c0

	if i, ok := ci.(sharedinformers.LegacyControllerInit); ok {
	i.Init(config, si, c.LookupAndReconcile)
	} else if i, ok := ci.(sharedinformers.ControllerInit); ok {
	i.Init(&sharedinformers.ControllerInitArgumentsImpl{si, config, c.LookupAndReconcile})
	}
	c.controller = uc
	queue.Reconcile = c.reconcile
	if c.Informers.WorkerQueues == nil {
	c.Informers.WorkerQueues = map[string]*controller.QueueWorker{}
	}
	c.Informers.WorkerQueues["MachineDeployment"] = queue

The init of this controller however registers already eventHandlers for machineSets which will subsequently try to put their owning machineDeployments in the not yet existing qeue:

cluster-api/pkg/controller/machinedeployment/controller.go

Lines 65 to 69 in 5b128c0

	arguments.GetSharedInformers().Factory.Cluster().V1alpha1().MachineSets().Informer().AddEventHandler(cache.ResourceEventHandlerFuncs{
	AddFunc: c.addMachineSet,
	UpdateFunc: c.updateMachineSet,
	DeleteFunc: c.deleteMachineSet,
	})

Just creating the qeue before calling Init is unfortunatelly not possible because all of that happens in generated code.

Since this should only happen during the startup in just a very small timeframe I think its okay like this. WDYT?

[roberthbailey]

@pwittrock - is this something that gets fixed with the kubebuilder change?

[Lion-Wei]

@alvaroaleman , I see, thanks.

[alvaroaleman]

@roberthbailey Are you fine with merging this to have the machineDeployment controller working again?

[roberthbailey]

Yes, sorry for the delay.

[roberthbailey]

/approve
/lgtm

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: alvaroaleman, roberthbailey

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

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [roberthbailey]

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