cluster-api-provider-aws: AWSMachineTemplate with spot instance generate false error log

/kind bug

What steps did you take and what happened: When I declare spotInstance on a new AWSMachineTemplate, like this:

--- a/clusters/infrastructure.cluster.x-k8s.io-v1beta1.AWSMachineTemplate-xxx-v1.yaml
+++ b/clusters/infrastructure.cluster.x-k8s.io-v1beta1.AWSMachineTemplate-xxx-v2.yaml
@@ -1,11 +1,13 @@
 apiVersion: infrastructure.cluster.x-k8s.io/v1beta1
 kind: AWSMachineTemplate
 metadata:
-  name: xxx-v1
+  name: xxx-v2
   namespace: flux-system
 spec:
   template:
     spec:
       iamInstanceProfile: nodes.cluster-api-provider-aws.sigs.k8s.io
       instanceType: xxx.medium
+      spotMarketOptions:
+        maxPrice: ""
       sshKeyName: xxx

Everything works fine. MachineDeployment using this template provision the new nodes, etc. They are seen by the management cluster and by the workload cluster.

But it generates a false error log on capi-controller-manager :

E0411 12:54:02.362276       1 controller.go:317] controller/machine "msg"="Reconciler error" "error"="machines.cluster.x-k8s.io \"XXXX\" not found" "name"=XXXX" "namespace"="flux-system" "reconciler group"="cluster.x-k8s.io" "reconciler kind"="Machine"

Even though the nodes and the machines are here, consistent, up and running.

What did you expect to happen:

I don’t get a false error log message when I use spot instances.

Anything else you would like to add:

When I remove spotMarketOptions, I don’t get this false error log.

Environment:

• Cluster-api version: 1.1.3
• Cluster-api-provider-aws version: 1.4.0
• Kubernetes version: (use kubectl version): v1.22.6-eks-7d68063
• OS (e.g. from /etc/os-release): EKS