Related to #692: Skip unsupported guest OS info calls

internal/monitoring/helpers_test.go

@@ -1,6 +1,7 @@
 package monitoring
 
 import (
+	"errors"
 	"fmt"
 	"math"
 	"testing"
@@ -278,3 +279,41 @@ func TestConvertPoolInfoToModelNil(t *testing.T) {
 		t.Fatalf("expected nil result for nil input")
 	}
 }
+
+func TestIsGuestAgentOSInfoUnsupportedError(t *testing.T) {
+	t.Parallel()
+
+	cases := []struct {
+		name string
+		err  error
+		want bool
+	}{
+		{name: "nil error", err: nil, want: false},
+		{name: "unrelated error", err: errors.New("guest agent timeout"), want: false},
+		{
+			name: "missing os-release path",
+			err:  errors.New(`API error 500: {"errors":{"message":"guest agent command failed: Failed to open file '/etc/os-release': No such file or directory"}}`),
+			want: true,
+		},
+		{
+			name: "missing usr lib os-release",
+			err:  errors.New("API error 500: guest agent command failed: Failed to open file '/usr/lib/os-release': No such file or directory"),
+			want: true,
+		},
+		{
+			name: "unsupported command",
+			err:  errors.New("API error 500: unsupported command: guest-get-osinfo"),
+			want: true,
+		},
+	}
+
+	for _, tc := range cases {
+		tc := tc
+		t.Run(tc.name, func(t *testing.T) {
+			t.Parallel()
+			if got := isGuestAgentOSInfoUnsupportedError(tc.err); got != tc.want {
+				t.Fatalf("isGuestAgentOSInfoUnsupportedError(%v) = %v, want %v", tc.err, got, tc.want)
+			}
+		})
+	}
+}

internal/monitoring/monitor.go

@@ -2485,6 +2485,16 @@ func (m *Monitor) fetchGuestAgentMetadata(ctx context.Context, client PVEClientI
 			return client.GetVMAgentInfo(ctx, nodeName, vmid)
 		})
 		if err != nil {
+			if isGuestAgentOSInfoUnsupportedError(err) {
+				osInfoSkip = true
+				osInfoFailureCount = guestAgentOSInfoFailureThreshold
+				log.Warn().
+					Str("instance", instanceName).
+					Str("vm", vmName).
+					Int("vmid", vmid).
+					Err(err).
+					Msg("Guest agent OS info unsupported (missing os-release). Skipping future calls to avoid qemu-ga issues (refs #692)")
+			} else {
 				osInfoFailureCount++
 				if osInfoFailureCount >= guestAgentOSInfoFailureThreshold {
 					osInfoSkip = true
@@ -2503,6 +2513,7 @@ func (m *Monitor) fetchGuestAgentMetadata(ctx context.Context, client PVEClientI
 						Err(err).
 						Msg("Guest agent OS info unavailable")
 				}
+			}
 		} else if agentInfo, ok := agentInfoRaw.(map[string]interface{}); ok && len(agentInfo) > 0 {
 			osName, osVersion = extractGuestOSInfo(agentInfo)
 			osInfoFailureCount = 0 // Reset on success
@@ -2720,6 +2731,27 @@ func extractGuestOSInfo(data map[string]interface{}) (string, string) {
 	return osName, osVersion
 }
 
+func isGuestAgentOSInfoUnsupportedError(err error) bool {
+	if err == nil {
+		return false
+	}
+
+	msg := strings.ToLower(err.Error())
+
+	// OpenBSD qemu-ga emits "Failed to open file '/etc/os-release'" (refs #692)
+	if strings.Contains(msg, "os-release") &&
+		(strings.Contains(msg, "failed to open file") || strings.Contains(msg, "no such file or directory")) {
+		return true
+	}
+
+	// Some Proxmox builds bubble up "unsupported command: guest-get-osinfo"
+	if strings.Contains(msg, "guest-get-osinfo") && strings.Contains(msg, "unsupported") {
+		return true
+	}
+
+	return false
+}
+
 func stringValue(val interface{}) string {
 	switch v := val.(type) {
 	case string: