feat: use proxmox ha-group as zone name

This feature enables live migration without changing any Kubernetes labels.

Signed-off-by: Serge Logvinov <serge.logvinov@sinextra.dev>

go.mod

@@ -6,7 +6,7 @@ require (
 	github.com/Telmate/proxmox-api-go v0.0.0-20250202141955-0f3daee49334
 	github.com/jarcoal/httpmock v1.4.1
 	github.com/pkg/errors v0.9.1
-	github.com/spf13/pflag v1.0.9
+	github.com/spf13/pflag v1.0.10
 	github.com/stretchr/testify v1.11.1
 	gopkg.in/yaml.v3 v3.0.1
 	k8s.io/api v0.34.0

go.sum

@@ -141,8 +141,9 @@ github.com/soheilhy/cmux v0.1.5 h1:jjzc5WVemNEDTLwv9tlmemhC73tI08BNOIGwBOo10Js=
 github.com/soheilhy/cmux v0.1.5/go.mod h1:T7TcVDs9LWfQgPlPsdngu6I6QIoyIFZDDC6sNE1GqG0=
 github.com/spf13/cobra v1.10.1 h1:lJeBwCfmrnXthfAupyUTzJ/J4Nc1RsHC/mSRU2dll/s=
 github.com/spf13/cobra v1.10.1/go.mod h1:7SmJGaTHFVBY0jW4NXGluQoLvhqFQM+6XSKD+P4XaB0=
-github.com/spf13/pflag v1.0.9 h1:9exaQaMOCwffKiiiYk6/BndUBv+iRViNW+4lEMi0PvY=
 github.com/spf13/pflag v1.0.9/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
+github.com/spf13/pflag v1.0.10 h1:4EBh2KAYBwaONj6b2Ye1GiHfwjqyROoF4RwYO+vPwFk=
+github.com/spf13/pflag v1.0.10/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 github.com/stoewer/go-strcase v1.3.1 h1:iS0MdW+kVTxgMoE1LAZyMiYJFKlOzLooE4MxjirtkAs=
 github.com/stoewer/go-strcase v1.3.1/go.mod h1:fAH5hQ5pehh+j3nZfvwdk2RgEgQjAoM8wodgtPmh1xo=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=

pkg/config/config.go

@@ -63,12 +63,23 @@ type NetworkOpts struct {
 	Mode                NetworkMode `yaml:"mode,omitempty"`
 }
 
+// ClustersFeatures specifies the features for the cloud provider.
+type ClustersFeatures struct {
+	// HAGroup specifies if the provider should use HA groups to determine node zone.
+	// If enabled, the provider will use the HA group name as the zone name.
+	// If disabled, the provider will use the node's cluster name as the zone name.
+	// Default is false.
+	HAGroup bool `yaml:"ha_group,omitempty"`
+	// Provider specifies the provider to use. Can be 'default' or 'capmox'.
+	// Default is 'default'.
+	Provider Provider `yaml:"provider,omitempty"`
+	// Network specifies the network options for the cloud provider.
+	Network NetworkOpts `yaml:"network,omitempty"`
+}
+
 // ClustersConfig is proxmox multi-cluster cloud config.
 type ClustersConfig struct {
-	Features struct {
+	Features ClustersFeatures              `yaml:"features,omitempty"`
-		Provider Provider    `yaml:"provider,omitempty"`
-		Network  NetworkOpts `yaml:"network,omitempty"`
-	} `yaml:"features,omitempty"`
 	Clusters []*proxmoxpool.ProxmoxCluster `yaml:"clusters,omitempty"`
 }
 

pkg/proxmox/cloud.go

@@ -74,7 +74,7 @@ func newCloud(config *ccmConfig.ClustersConfig) (cloudprovider.Interface, error)
 		return nil, err
 	}
 
-	instancesInterface := newInstances(client, config.Features.Provider, config.Features.Network)
+	instancesInterface := newInstances(client, config.Features)
 
 	return &cloud{
 		client:      client,

pkg/proxmox/instances.go

@@ -55,44 +55,42 @@ type instanceInfo struct {
 }
 
 type instances struct {
-	c           *client
+	c             *client
-	provider    providerconfig.Provider
+	zoneAsHAGroup bool
-	networkOpts instanceNetops
+	provider      providerconfig.Provider
+	networkOpts   instanceNetops
 }
 
 var instanceTypeNameRegexp = regexp.MustCompile(`(^[a-zA-Z0-9_.-]+)$`)
 
-func newInstances(
+func newInstances(client *client, features providerconfig.ClustersFeatures) *instances {
-	client *client,
+	externalIPCIDRs := ParseCIDRList(features.Network.ExternalIPCIDRS)
-	provider providerconfig.Provider,
+	if len(features.Network.ExternalIPCIDRS) > 0 && len(externalIPCIDRs) == 0 {
-	networkOpts providerconfig.NetworkOpts,
+		klog.Warningf("Failed to parse external CIDRs: %v", features.Network.ExternalIPCIDRS)
-) *instances {
-	externalIPCIDRs := ParseCIDRList(networkOpts.ExternalIPCIDRS)
-	if len(networkOpts.ExternalIPCIDRS) > 0 && len(externalIPCIDRs) == 0 {
-		klog.Warningf("Failed to parse external CIDRs: %v", networkOpts.ExternalIPCIDRS)
 	}
 
-	sortOrderCIDRs, ignoredCIDRs, err := ParseCIDRRuleset(networkOpts.IPSortOrder)
+	sortOrderCIDRs, ignoredCIDRs, err := ParseCIDRRuleset(features.Network.IPSortOrder)
 	if err != nil {
 		klog.Errorf("Failed to parse sort order CIDRs: %v", err)
 	}
 
-	if len(networkOpts.IPSortOrder) > 0 && (len(sortOrderCIDRs)+len(ignoredCIDRs)) == 0 {
+	if len(features.Network.IPSortOrder) > 0 && (len(sortOrderCIDRs)+len(ignoredCIDRs)) == 0 {
-		klog.Warningf("Failed to parse sort order CIDRs: %v", networkOpts.IPSortOrder)
+		klog.Warningf("Failed to parse sort order CIDRs: %v", features.Network.IPSortOrder)
 	}
 
 	netOps := instanceNetops{
 		ExternalCIDRs:       externalIPCIDRs,
 		SortOrder:           sortOrderCIDRs,
 		IgnoredCIDRs:        ignoredCIDRs,
-		Mode:                networkOpts.Mode,
+		Mode:                features.Network.Mode,
-		IPv6SupportDisabled: networkOpts.IPv6SupportDisabled,
+		IPv6SupportDisabled: features.Network.IPv6SupportDisabled,
 	}
 
 	return &instances{
-		c:           client,
+		c:             client,
-		provider:    provider,
+		zoneAsHAGroup: features.HAGroup,
-		networkOpts: netOps,
+		provider:      features.Provider,
+		networkOpts:   netOps,
 	}
 }
 
@@ -224,12 +222,6 @@ func (i *instances) InstanceMetadata(ctx context.Context, node *v1.Node) (*cloud
 		}
 	}
 
-	if len(additionalLabels) > 0 && !hasUninitializedTaint(node) {
-		if err := syncNodeLabels(i.c, node, additionalLabels); err != nil {
-			klog.ErrorS(err, "error updating labels for the node", "node", klog.KRef("", node.Name))
-		}
-	}
-
 	metadata := &cloudprovider.InstanceMetadata{
 		ProviderID:       providerID,
 		NodeAddresses:    i.addresses(ctx, node, info),
@@ -239,6 +231,24 @@ func (i *instances) InstanceMetadata(ctx context.Context, node *v1.Node) (*cloud
 		AdditionalLabels: additionalLabels,
 	}
 
+	if i.zoneAsHAGroup {
+		haGroup, err := i.c.pxpool.GetNodeGroup(ctx, info.Region, info.Node)
+		if err != nil {
+			klog.ErrorS(err, "instances.InstanceMetadata() failed to get HA group for the node", "node", klog.KRef("", node.Name), "region", info.Region)
+
+			return nil, err
+		}
+
+		metadata.Zone = haGroup
+		additionalLabels[LabelTopologyHAGroup] = haGroup
+	}
+
+	if len(additionalLabels) > 0 && !hasUninitializedTaint(node) {
+		if err := syncNodeLabels(i.c, node, additionalLabels); err != nil {
+			klog.ErrorS(err, "error updating labels for the node", "node", klog.KRef("", node.Name))
+		}
+	}
+
 	klog.V(5).InfoS("instances.InstanceMetadata()", "info", info, "metadata", metadata)
 
 	return metadata, nil

pkg/proxmox/instances_test.go

@@ -179,7 +179,12 @@ clusters:
 		kclient: fake.NewSimpleClientset(),
 	}
 
-	ts.i = newInstances(client, providerconfig.ProviderDefault, providerconfig.NetworkOpts{})
+	features := providerconfig.ClustersFeatures{
+		Provider: providerconfig.ProviderDefault,
+		Network:  providerconfig.NetworkOpts{},
+	}
+
+	ts.i = newInstances(client, features)
 }
 
 func (ts *ccmTestSuite) TearDownTest() {

pkg/proxmoxpool/pool.go

@@ -143,12 +143,7 @@ func (c *ProxmoxPool) GetProxmoxCluster(region string) (*proxmox.Client, error)
 
 // GetNodeGroup returns a Proxmox node ha-group in a given region.
 func (c *ProxmoxPool) GetNodeGroup(ctx context.Context, region string, node string) (string, error) {
-	px, err := c.GetProxmoxCluster(region)
+	haGroups, err := c.GetHAGroupList(ctx, region)
-	if err != nil {
-		return "", err
-	}
-
-	haGroups, err := px.GetHAGroupList(ctx)
 	if err != nil {
 		return "", fmt.Errorf("error get ha-groups %v", err)
 	}
@@ -168,6 +163,44 @@ func (c *ProxmoxPool) GetNodeGroup(ctx context.Context, region string, node stri
 	return "", ErrHAGroupNotFound
 }
 
+// GetHAGroupList returns a list of Proxmox ha-groups in a given region.
+func (c *ProxmoxPool) GetHAGroupList(ctx context.Context, region string) (haGroups []proxmox.HAGroup, err error) {
+	px, err := c.GetProxmoxCluster(region)
+	if err != nil {
+		return nil, err
+	}
+
+	list, err := px.GetItemList(ctx, "/cluster/ha/groups")
+	if err != nil {
+		return nil, err
+	}
+
+	haGroups = []proxmox.HAGroup{}
+
+	items, ok := list["data"].([]interface{})
+	if !ok {
+		return nil, fmt.Errorf("failed to cast response to list of HA groups in region %s, error: %v", region, err)
+	}
+
+	for _, item := range items {
+		itemMap := item.(map[string]interface{})
+
+		if itemMap["type"].(string) != "group" {
+			continue
+		}
+
+		haGroups = append(haGroups, proxmox.HAGroup{
+			Group:      itemMap["group"].(string),
+			Nodes:      strings.Split(itemMap["nodes"].(string), ","),
+			NoFailback: itemMap["nofailback"].(float64) == 1,
+			Restricted: itemMap["restricted"].(float64) == 1,
+			Type:       itemMap["type"].(string),
+		})
+	}
+
+	return haGroups, nil
+}
+
 // FindVMByNode find a VM by kubernetes node resource in all Proxmox clusters.
 func (c *ProxmoxPool) FindVMByNode(ctx context.Context, node *v1.Node) (*proxmox.VmRef, string, error) {
 	for region, px := range c.clients {