1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
|
package kubernetes
import (
"errors"
"fmt"
"io"
"net/http"
"strings"
"time"
"golang.org/x/net/context"
api "k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/api/resource"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/client-go/kubernetes"
restclient "k8s.io/client-go/rest"
"k8s.io/client-go/tools/clientcmd"
"gitlab.com/gitlab-org/gitlab-runner/common"
"gitlab.com/gitlab-org/gitlab-runner/executors/kubernetes/internal/pull"
)
type kubeConfigProvider func() (*restclient.Config, error)
type resourceQuantityError struct {
resource string
value string
inner error
}
func (r *resourceQuantityError) Error() string {
return fmt.Sprintf("parsing resource %q with value %q: %q", r.resource, r.value, r.inner)
}
func (r *resourceQuantityError) Is(err error) bool {
t, ok := err.(*resourceQuantityError)
return ok && r.resource == t.resource && r.value == t.value && r.inner == t.inner
}
var (
// inClusterConfig parses kubernetes configuration reading in cluster values
inClusterConfig kubeConfigProvider = restclient.InClusterConfig
// defaultKubectlConfig parses kubectl configuration ad loads the default cluster
defaultKubectlConfig kubeConfigProvider = loadDefaultKubectlConfig
)
func getKubeClientConfig(
config *common.KubernetesConfig,
overwrites *overwrites,
) (kubeConfig *restclient.Config, err error) {
if len(config.Host) > 0 {
kubeConfig, err = getOutClusterClientConfig(config)
} else {
kubeConfig, err = guessClientConfig()
}
if err != nil {
return nil, err
}
// apply overwrites
if overwrites.bearerToken != "" {
kubeConfig.BearerToken = overwrites.bearerToken
}
kubeConfig.UserAgent = common.AppVersion.UserAgent()
return kubeConfig, nil
}
func getOutClusterClientConfig(config *common.KubernetesConfig) (*restclient.Config, error) {
kubeConfig := &restclient.Config{
Host: config.Host,
BearerToken: config.BearerToken,
TLSClientConfig: restclient.TLSClientConfig{
CAFile: config.CAFile,
},
}
// certificate based auth
if config.CertFile != "" {
if config.KeyFile == "" || config.CAFile == "" {
return nil, fmt.Errorf("ca file, cert file and key file must be specified when using file based auth")
}
kubeConfig.TLSClientConfig.CertFile = config.CertFile
kubeConfig.TLSClientConfig.KeyFile = config.KeyFile
}
return kubeConfig, nil
}
func guessClientConfig() (*restclient.Config, error) {
// Try in cluster config first
if inClusterCfg, err := inClusterConfig(); err == nil {
return inClusterCfg, nil
}
// in cluster config failed. Reading default kubectl config
return defaultKubectlConfig()
}
func loadDefaultKubectlConfig() (*restclient.Config, error) {
config, err := clientcmd.NewDefaultClientConfigLoadingRules().Load()
if err != nil {
return nil, err
}
return clientcmd.NewDefaultClientConfig(*config, &clientcmd.ConfigOverrides{}).ClientConfig()
}
func closeKubeClient(client *kubernetes.Clientset) bool {
if client == nil {
return false
}
rest, ok := client.CoreV1().RESTClient().(*restclient.RESTClient)
if !ok || rest.Client == nil || rest.Client.Transport == nil {
return false
}
if transport, ok := rest.Client.Transport.(*http.Transport); ok {
transport.CloseIdleConnections()
return true
}
return false
}
func isRunning(pod *api.Pod) (bool, error) {
switch pod.Status.Phase {
case api.PodRunning:
return true, nil
case api.PodSucceeded:
return false, fmt.Errorf("pod already succeeded before it begins running")
case api.PodFailed:
return false, fmt.Errorf("pod status is failed")
default:
return false, nil
}
}
type podPhaseResponse struct {
done bool
phase api.PodPhase
err error
}
func getPodPhase(c *kubernetes.Clientset, pod *api.Pod, out io.Writer) podPhaseResponse {
// TODO: handle the context properly with https://gitlab.com/gitlab-org/gitlab-runner/-/issues/27932
pod, err := c.CoreV1().Pods(pod.Namespace).Get(context.TODO(), pod.Name, metav1.GetOptions{})
if err != nil {
return podPhaseResponse{true, api.PodUnknown, err}
}
ready, err := isRunning(pod)
if err != nil || ready {
return podPhaseResponse{true, pod.Status.Phase, err}
}
// check status of containers
for _, container := range append(pod.Status.ContainerStatuses, pod.Status.InitContainerStatuses...) {
if container.Ready {
continue
}
waiting := container.State.Waiting
if waiting == nil {
continue
}
switch waiting.Reason {
case "InvalidImageName":
err = &common.BuildError{Inner: fmt.Errorf("image pull failed: %s", waiting.Message)}
return podPhaseResponse{true, api.PodUnknown, err}
case "ErrImagePull", "ImagePullBackOff":
msg := fmt.Sprintf("image pull failed: %s", waiting.Message)
imagePullErr := &pull.ImagePullError{Message: msg, Image: container.Image}
return podPhaseResponse{
true,
api.PodUnknown,
&common.BuildError{Inner: imagePullErr, FailureReason: common.ScriptFailure},
}
}
}
_, _ = fmt.Fprintf(
out,
"Waiting for pod %s/%s to be running, status is %s\n",
pod.Namespace,
pod.Name,
pod.Status.Phase,
)
for _, condition := range pod.Status.Conditions {
// skip conditions with no reason, these are typically expected pod conditions
if condition.Reason == "" {
continue
}
_, _ = fmt.Fprintf(
out,
"\t%s: %q\n",
condition.Reason,
condition.Message,
)
}
return podPhaseResponse{false, pod.Status.Phase, nil}
}
func triggerPodPhaseCheck(c *kubernetes.Clientset, pod *api.Pod, out io.Writer) <-chan podPhaseResponse {
errc := make(chan podPhaseResponse)
go func() {
defer close(errc)
errc <- getPodPhase(c, pod, out)
}()
return errc
}
// waitForPodRunning will use client c to detect when pod reaches the PodRunning
// state. It returns the final PodPhase once either PodRunning, PodSucceeded or
// PodFailed has been reached. In the case of PodRunning, it will also wait until
// all containers within the pod are also Ready.
// It returns error if the call to retrieve pod details fails or the timeout is
// reached.
// The timeout and polling values are configurable through KubernetesConfig
// parameters.
func waitForPodRunning(
ctx context.Context,
c *kubernetes.Clientset,
pod *api.Pod,
out io.Writer,
config *common.KubernetesConfig,
) (api.PodPhase, error) {
pollInterval := config.GetPollInterval()
pollAttempts := config.GetPollAttempts()
for i := 0; i <= pollAttempts; i++ {
select {
case r := <-triggerPodPhaseCheck(c, pod, out):
if !r.done {
time.Sleep(time.Duration(pollInterval) * time.Second)
continue
}
return r.phase, r.err
case <-ctx.Done():
return api.PodUnknown, ctx.Err()
}
}
return api.PodUnknown, errors.New("timed out waiting for pod to start")
}
// limits takes a string representing CPU, memory and ephemeralStorage limits,
// and returns a ResourceList with appropriately scaled Quantity
// values for Kubernetes. This allows users to write "500m" for CPU,
// "50Mi" for memory and "1Gi" for ephemeral storage (etc.)
func createResourceList(cpu, memory, ephemeralStorage string) (api.ResourceList, error) {
var rCPU, rMem, rStor resource.Quantity
var err error
parse := func(s string) (resource.Quantity, error) {
var q resource.Quantity
if s == "" {
return q, nil
}
if q, err = resource.ParseQuantity(s); err != nil {
return q, err
}
return q, nil
}
if rCPU, err = parse(cpu); err != nil {
return api.ResourceList{}, &resourceQuantityError{resource: "cpu", value: cpu, inner: err}
}
if rMem, err = parse(memory); err != nil {
return api.ResourceList{}, &resourceQuantityError{resource: "memory", value: memory, inner: err}
}
if rStor, err = parse(ephemeralStorage); err != nil {
return api.ResourceList{}, &resourceQuantityError{
resource: "ephemeralStorage",
value: ephemeralStorage,
inner: err,
}
}
l := make(api.ResourceList)
q := resource.Quantity{}
if rCPU != q {
l[api.ResourceCPU] = rCPU
}
if rMem != q {
l[api.ResourceMemory] = rMem
}
if rStor != q {
l[api.ResourceEphemeralStorage] = rStor
}
return l, nil
}
// buildVariables converts a common.BuildVariables into a list of
// kubernetes EnvVar objects
func buildVariables(bv common.JobVariables) []api.EnvVar {
e := make([]api.EnvVar, len(bv))
for i, b := range bv {
e[i] = api.EnvVar{
Name: b.Key,
Value: b.Value,
}
}
return e
}
// Sanitize labels to match Kubernetes restrictions from https://kubernetes.io/
// /docs/concepts/overview/working-with-objects/labels/#syntax-and-character-set
//nolint:gocognit
func sanitizeLabel(value string) string {
mapFn := func(r rune) rune {
if r >= 'a' && r <= 'z' ||
r >= 'A' && r <= 'Z' ||
r >= '0' && r <= '9' ||
r == '-' || r == '_' || r == '.' {
return r
}
return '_'
}
// only alphanumerics, dashes (-), underscores (_), dots (.) are valid
value = strings.Map(mapFn, value)
// must start/end with alphanumerics only
value = strings.Trim(value, "-_.")
// length must be <= 63 characters
if len(value) > 63 {
value = value[:63]
}
// trim again if required after shortening
return strings.Trim(value, "-_.")
}
|