Replace pid with flock for runtime config loading

Use lock file and flock(2) to ensure there is only a single instance of
k0s running. This is more reliable than storing the pid in the runtime
config.

This also solves false positives with k0s runtime config leftovers.

Fixes: #5399
Signed-off-by: Natanael Copa <[email protected]>

go.mod

@@ -22,6 +22,7 @@ require (
 	github.com/go-logr/logr v1.4.2
 	github.com/go-openapi/jsonpointer v0.21.0
 	github.com/go-playground/validator/v10 v10.24.0
+	github.com/gofrs/flock v0.8.1
 	github.com/google/go-cmp v0.6.0
 	github.com/k0sproject/bootloose v0.9.0
 	github.com/k0sproject/version v0.6.0

go.sum

@@ -235,6 +235,8 @@ github.com/gobwas/glob v0.2.3/go.mod h1:d3Ez4x06l9bZtSvzIay5+Yzi0fmZzPgnTbPcKjJA
 github.com/godbus/dbus/v5 v5.0.4/go.mod h1:xhWf0FNVPg57R7Z0UbKHbJfkEywrmjJnf7w5xrFpKfA=
 github.com/godbus/dbus/v5 v5.1.0 h1:4KLkAxT3aOY8Li4FRJe/KvhoNFFxo0m6fNuFUO8QJUk=
 github.com/godbus/dbus/v5 v5.1.0/go.mod h1:xhWf0FNVPg57R7Z0UbKHbJfkEywrmjJnf7w5xrFpKfA=
+github.com/gofrs/flock v0.8.1 h1:+gYjHKf32LDeiEEFhQaotPbLuUXjY5ZqxKgXy7n59aw=
+github.com/gofrs/flock v0.8.1/go.mod h1:F1TvTiK9OcQqauNUHlbJvyl9Qa1QvF/gOUDKA14jxHU=
 github.com/gogo/protobuf v1.1.1/go.mod h1:r8qH/GZQm5c6nD/R0oafs1akxWv10x8SbQlK7atdtwQ=
 github.com/gogo/protobuf v1.3.2 h1:Ov1cvc58UF3b5XjBnZv7+opcTcQFZebYjWzi34vdm4Q=
 github.com/gogo/protobuf v1.3.2/go.mod h1:P1XiOD3dCwIKUDQYPy72D8LYyHL2YPYrpS2s69NZV8Q=

pkg/config/runtime.go

@@ -22,6 +22,7 @@ import (
 	"os"
 	"path/filepath"
 
+	"github.com/gofrs/flock"
 	"github.com/k0sproject/k0s/internal/pkg/dir"
 	"github.com/k0sproject/k0s/pkg/apis/k0s/v1beta1"
 	"github.com/k0sproject/k0s/pkg/constant"
@@ -56,7 +57,7 @@ type RuntimeConfig struct {
 type RuntimeConfigSpec struct {
 	NodeConfig *v1beta1.ClusterConfig `json:"nodeConfig"`
 	K0sVars    *CfgVars               `json:"k0sVars"`
-	Pid        int                    `json:"pid"`
+	lock       *flock.Flock
 }
 
 func LoadRuntimeConfig(k0sVars *CfgVars) (*RuntimeConfigSpec, error) {
@@ -83,22 +84,20 @@ func LoadRuntimeConfig(k0sVars *CfgVars) (*RuntimeConfigSpec, error) {
 		return nil, fmt.Errorf("%w: spec is nil", ErrInvalidRuntimeConfig)
 	}
 
-	// If a pid is defined but there's no process found, the instance of k0s is
-	// expected to have died, in which case the existing config is removed and
-	// an error is returned, which allows the controller startup to proceed to
-	// initialize a new runtime config.
-	if spec.Pid != 0 {
-		if err := checkPid(spec.Pid); err != nil {
-			defer func() { _ = spec.Cleanup() }()
-			return nil, errors.Join(ErrK0sNotRunning, err)
-		}
-	}
-
 	return spec, nil
 }
 
 func NewRuntimeConfig(k0sVars *CfgVars) (*RuntimeConfigSpec, error) {
-	if _, err := LoadRuntimeConfig(k0sVars); err == nil {
+	if err := dir.Init(filepath.Dir(k0sVars.RuntimeConfigPath), constant.RunDirMode); err != nil {
+		logrus.Warnf("failed to initialize runtime config dir: %v", err)
+	}
+
+	lock := flock.New(k0sVars.RuntimeConfigPath + ".lock")
+	locked, err := lock.TryLock()
+	if err != nil {
+		return nil, fmt.Errorf("failed to aquire lock on runtime config: %w", err)
+	}
+	if !locked {
 		return nil, ErrK0sAlreadyRunning
 	}
 
@@ -123,7 +122,7 @@ func NewRuntimeConfig(k0sVars *CfgVars) (*RuntimeConfigSpec, error) {
 		Spec: &RuntimeConfigSpec{
 			NodeConfig: nodeConfig,
 			K0sVars:    k0sVars,
-			Pid:        os.Getpid(),
+			lock:       lock,
 		},
 	}
 
@@ -132,10 +131,6 @@ func NewRuntimeConfig(k0sVars *CfgVars) (*RuntimeConfigSpec, error) {
 		return nil, err
 	}
 
-	if err := dir.Init(filepath.Dir(k0sVars.RuntimeConfigPath), constant.RunDirMode); err != nil {
-		logrus.Warnf("failed to initialize runtime config dir: %v", err)
-	}
-
 	if err := os.WriteFile(k0sVars.RuntimeConfigPath, content, 0600); err != nil {
 		return nil, fmt.Errorf("failed to write runtime config: %w", err)
 	}
@@ -149,7 +144,15 @@ func (r *RuntimeConfigSpec) Cleanup() error {
 	}
 
 	if err := os.Remove(r.K0sVars.RuntimeConfigPath); err != nil {
-		return fmt.Errorf("failed to clean up runtime config file: %w", err)
+		logrus.Warnf("failed to clean up runtime config file: %v", err)
+	}
+
+	if err := r.lock.Close(); err != nil {
+		return fmt.Errorf("failed to unlock runtime config: %w", err)
+	}
+
+	if err := os.Remove(r.lock.Path()); err != nil {
+		return fmt.Errorf("failed to delete %s: %w", r.lock.Path(), err)
 	}
 	return nil
 }

pkg/config/runtime_test.go

@@ -22,12 +22,13 @@ import (
 
 	"github.com/k0sproject/k0s/pkg/apis/k0s/v1beta1"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
-func TestLoadRuntimeConfig_K0sNotRunning(t *testing.T) {
+func TestLoadRuntimeConfig(t *testing.T) {
 	// create a temporary file for runtime config
 	tmpfile, err := os.CreateTemp("", "runtime-config")
-	assert.NoError(t, err)
+	require.NoError(t, err)
 	defer os.Remove(tmpfile.Name())
 
 	// prepare k0sVars
@@ -43,26 +44,29 @@ spec:
   nodeConfig:
     metadata:
       name: k0s
-  pid: 9999999
 `)
 	err = os.WriteFile(k0sVars.RuntimeConfigPath, content, 0644)
-	assert.NoError(t, err)
+	require.NoError(t, err)
 
 	// try to load runtime config and check if it returns an error
 	spec, err := LoadRuntimeConfig(k0sVars)
-	assert.Nil(t, spec)
-	assert.ErrorIs(t, err, ErrK0sNotRunning)
+	require.NoError(t, err)
+	assert.NotNil(t, spec)
 }
 
 func TestNewRuntimeConfig(t *testing.T) {
 	// create a temporary directory for k0s files
 	tempDir, err := os.MkdirTemp("", "k0s")
-	assert.NoError(t, err)
+	require.NoError(t, err)
 	defer os.RemoveAll(tempDir)
 
+	tempRunDir, err := os.MkdirTemp("", "k0s-rundir")
+	require.NoError(t, err)
+	defer os.RemoveAll(tempRunDir)
+
 	// create a temporary file for the runtime config
-	tmpfile, err := os.CreateTemp("", "runtime-config")
-	assert.NoError(t, err)
+	tmpfile, err := os.CreateTemp(tempRunDir, "runtime-config")
+	require.NoError(t, err)
 	tmpfile.Close()
 	defer os.Remove(tmpfile.Name())
 
@@ -82,7 +86,6 @@ func TestNewRuntimeConfig(t *testing.T) {
 	assert.NoError(t, err)
 	assert.NotNil(t, spec)
 	assert.Equal(t, tempDir, spec.K0sVars.DataDir)
-	assert.Equal(t, os.Getpid(), spec.Pid)
 	assert.NotNil(t, spec.NodeConfig)
 	cfg, err := spec.K0sVars.NodeConfig()
 	assert.NoError(t, err)