feat: find node by uuid

In some setups, the Proxmox VM name may differ from the Linux hostname.
To reliably identify a VM within a Proxmox cluster, we can use the system's UUID

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

go.sum

@@ -135,8 +135,6 @@ github.com/prometheus/client_golang v1.20.3 h1:oPksm4K8B+Vt35tUhw6GbSNSgVlVSBH0q
 github.com/prometheus/client_golang v1.20.3/go.mod h1:PIEt8X02hGcP8JWbeHyeZ53Y/jReSnHgO035n//V5WE=
 github.com/prometheus/client_model v0.6.1 h1:ZKSh/rekM+n3CeS952MLRAdFwIKqeY8b62p8ais2e9E=
 github.com/prometheus/client_model v0.6.1/go.mod h1:OrxVMOVHjw3lKMa8+x6HeMGkHMQyHDk9E3jmP2AmGiY=
-github.com/prometheus/common v0.55.0 h1:KEi6DK7lXW/m7Ig5i47x0vRzuBsHuvJdi5ee6Y3G1dc=
-github.com/prometheus/common v0.55.0/go.mod h1:2SECS4xJG1kd8XF9IcM1gMX6510RAEL65zxzNImwdc8=
 github.com/prometheus/common v0.59.1 h1:LXb1quJHWm1P6wq/U824uxYi4Sg0oGvNeUm1z5dJoX0=
 github.com/prometheus/common v0.59.1/go.mod h1:GpWM7dewqmVYcd7SmRaiWVe9SSqjf0UrwnYnpEZNuT0=
 github.com/prometheus/procfs v0.15.1 h1:YagwOFzUgYfKKHX6Dr+sHT7km/hxC76UB0learggepc=
@@ -216,8 +214,6 @@ golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8U
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.27.0 h1:GXm2NjJrPaiv/h1tb2UH8QfgC/hOf/+z0p6PT8o1w7A=
 golang.org/x/crypto v0.27.0/go.mod h1:1Xngt8kV6Dvbssa53Ziq6Eqn0HqbZi5Z6R0ZpwQzt70=
-golang.org/x/exp v0.0.0-20240808152545-0cdaa3abc0fa h1:ELnwvuAXPNtPk1TJRuGkI9fDTwym6AYBu0qzT8AcHdI=
-golang.org/x/exp v0.0.0-20240808152545-0cdaa3abc0fa/go.mod h1:akd2r19cwCdwSwWeIdzYQGa/EZZyqcOdwWiwj5L5eKQ=
 golang.org/x/exp v0.0.0-20240909161429-701f63a606c0 h1:e66Fs6Z+fZTbFBAxKfP3PALWBtpfqks2bwGcexMxgtk=
 golang.org/x/exp v0.0.0-20240909161429-701f63a606c0/go.mod h1:2TbTHSBQa924w8M6Xs1QcRcFwyucIwBGpK1p2f1YFFY=
 golang.org/x/mod v0.2.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
@@ -228,8 +224,6 @@ golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLL
 golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.29.0 h1:5ORfpBpCs4HzDYoodCDBbwHzdR5UrLBZ3sOnUJmFoHo=
 golang.org/x/net v0.29.0/go.mod h1:gLkgy8jTGERgjzMic6DS9+SP0ajcu6Xu3Orq/SpETg0=
-golang.org/x/oauth2 v0.22.0 h1:BzDx2FehcG7jJwgWLELCdmLuxk2i+x9UDpSiss2u0ZA=
-golang.org/x/oauth2 v0.22.0/go.mod h1:XYTD2NtWslqkgxebSiOHnXEap4TF09sJSc7H1sXbhtI=
 golang.org/x/oauth2 v0.23.0 h1:PbgcYx2W7i4LvjJWEbf0ngHV6qJYr86PkAV3bXdLEbs=
 golang.org/x/oauth2 v0.23.0/go.mod h1:XYTD2NtWslqkgxebSiOHnXEap4TF09sJSc7H1sXbhtI=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
@@ -255,8 +249,8 @@ golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGm
 golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20200619180055-7c47624df98f/go.mod h1:EkVYQZoAsY45+roYkvgYkIh4xh/qjgUK9TdY2XT94GE=
 golang.org/x/tools v0.0.0-20210106214847-113979e3529a/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
-golang.org/x/tools v0.24.0 h1:J1shsA93PJUEVaUSaay7UXAyE8aimq3GW0pjlolpa24=
+golang.org/x/tools v0.25.0 h1:oFU9pkj/iJgs+0DT+VMHrx+oBKs/LJMV+Uvg78sl+fE=
-golang.org/x/tools v0.24.0/go.mod h1:YhNqVBIfWHdzvTLs0d8LCuMhkKUgSUKldakyV7W/WDQ=
+golang.org/x/tools v0.25.0/go.mod h1:/vtpO8WL1N9cQC3FN5zPqb//fRXskFHbLKk4OW1Q7rg=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=

pkg/cluster/client.go

@@ -19,8 +19,10 @@ package cluster
 
 import (
 	"crypto/tls"
+	"encoding/base64"
 	"fmt"
 	"net/http"
+	"net/url"
 	"os"
 	"strings"
 
@@ -107,3 +109,67 @@ func (c *Cluster) FindVMByName(name string) (*pxapi.VmRef, string, error) {
 
 	return nil, "", fmt.Errorf("vm '%s' not found", name)
 }
+
+// FindVMByUUID find a VM by uuid in all Proxmox clusters.
+func (c *Cluster) FindVMByUUID(uuid string) (*pxapi.VmRef, string, error) {
+	for region, px := range c.proxmox {
+		vms, err := px.GetResourceList("vm")
+		if err != nil {
+			return nil, "", fmt.Errorf("error get resources %v", err)
+		}
+
+		for vmii := range vms {
+			vm, ok := vms[vmii].(map[string]interface{})
+			if !ok {
+				return nil, "", fmt.Errorf("failed to cast response to map, vm: %v", vm)
+			}
+
+			if vm["type"].(string) != "qemu" {
+				continue
+			}
+
+			vmr := pxapi.NewVmRef(int(vm["vmid"].(float64)))
+			vmr.SetNode(vm["node"].(string))
+			vmr.SetVmType("qemu")
+
+			config, err := px.GetVmConfig(vmr)
+			if err != nil {
+				return nil, "", err
+			}
+
+			if config["smbios1"] != nil {
+				if c.getUUID(config["smbios1"].(string)) == uuid {
+					return vmr, region, nil
+				}
+			}
+		}
+	}
+
+	return nil, "", fmt.Errorf("vm with uuid '%s' not found", uuid)
+}
+
+func (c *Cluster) getUUID(smbios string) string {
+	for _, l := range strings.Split(smbios, ",") {
+		if l == "" || l == "base64=1" {
+			continue
+		}
+
+		parsedParameter, err := url.ParseQuery(l)
+		if err != nil {
+			return ""
+		}
+
+		for k, v := range parsedParameter {
+			if k == "uuid" {
+				decodedString, err := base64.StdEncoding.DecodeString(v[0])
+				if err != nil {
+					decodedString = []byte(v[0])
+				}
+
+				return string(decodedString)
+			}
+		}
+	}
+
+	return ""
+}

pkg/proxmox/instances.go

@@ -48,6 +48,12 @@ func newInstances(client *cluster.Cluster) *instances {
 func (i *instances) InstanceExists(_ context.Context, node *v1.Node) (bool, error) {
 	klog.V(4).InfoS("instances.InstanceExists() called", "node", klog.KRef("", node.Name))
 
+	if node.Spec.ProviderID == "" {
+		klog.V(4).InfoS("instances.InstanceExists() empty providerID, omitting unmanaged node", "node", klog.KObj(node))
+
+		return true, nil
+	}
+
 	if !strings.HasPrefix(node.Spec.ProviderID, provider.ProviderName) {
 		klog.V(4).InfoS("instances.InstanceExists() omitting unmanaged node", "node", klog.KObj(node), "providerID", node.Spec.ProviderID)
 
@@ -73,6 +79,12 @@ func (i *instances) InstanceExists(_ context.Context, node *v1.Node) (bool, erro
 func (i *instances) InstanceShutdown(_ context.Context, node *v1.Node) (bool, error) {
 	klog.V(4).InfoS("instances.InstanceShutdown() called", "node", klog.KRef("", node.Name))
 
+	if node.Spec.ProviderID == "" {
+		klog.V(4).InfoS("instances.InstanceShutdown() empty providerID, omitting unmanaged node", "node", klog.KObj(node))
+
+		return false, nil
+	}
+
 	if !strings.HasPrefix(node.Spec.ProviderID, provider.ProviderName) {
 		klog.V(4).InfoS("instances.InstanceShutdown() omitting unmanaged node", "node", klog.KObj(node), "providerID", node.Spec.ProviderID)
 
@@ -122,13 +134,20 @@ func (i *instances) InstanceMetadata(_ context.Context, node *v1.Node) (*cloudpr
 
 		providerID := node.Spec.ProviderID
 		if providerID == "" {
-			klog.V(4).InfoS("instances.InstanceMetadata() empty providerID, trying find by Name", "node", klog.KObj(node))
+			uuid := node.Status.NodeInfo.SystemUUID
+
+			klog.V(4).InfoS("instances.InstanceMetadata() empty providerID, trying find node", "node", klog.KObj(node), "uuid", uuid)
 
 			mc := metrics.NewMetricContext("findVmByName")
 
 			vmRef, region, err = i.c.FindVMByName(node.Name)
 			if mc.ObserveRequest(err) != nil {
-				return nil, fmt.Errorf("instances.InstanceMetadata() - failed to find instance by name %s: %v, skipped", node.Name, err)
+				mc := metrics.NewMetricContext("findVmByUUID")
+
+				vmRef, region, err = i.c.FindVMByUUID(uuid)
+				if mc.ObserveRequest(err) != nil {
+					return nil, fmt.Errorf("instances.InstanceMetadata() - failed to find instance by name/uuid %s: %v, skipped", node.Name, err)
+				}
 			}
 		} else if !strings.HasPrefix(node.Spec.ProviderID, provider.ProviderName) {
 			klog.V(4).InfoS("instances.InstanceMetadata() omitting unmanaged node", "node", klog.KObj(node), "providerID", node.Spec.ProviderID)