mirror of
https://github.com/k3s-io/kubernetes.git
synced 2025-07-22 11:21:47 +00:00
Merge pull request #130726 from rata/userns-e2e-kubelet-mappings
Userns: Add e2e tests for custom kubelet mappings, skip on windows and minor improvements
This commit is contained in:
commit
1704c52a2e
@ -131,6 +131,12 @@ func (kl *Kubelet) getKubeletMappings() (uint32, uint32, error) {
|
||||
return defaultFirstID, defaultLen, nil
|
||||
}
|
||||
|
||||
// We NEED to check for the user because getsubids can be configured to gather the response
|
||||
// with a remote call and we can't distinguish between the remote endpoint not being reachable
|
||||
// and the remote endpoint is reachable but no entry is present for the user.
|
||||
// So we check for the kubelet user first, if it exist and getsubids is present, we expect
|
||||
// to get _some_ configuration. If the user exist and getsubids doesn't give us any
|
||||
// configuration, then we consider the remote down and fail to start the kubelet.
|
||||
_, err := user.Lookup(kubeletUser)
|
||||
if err != nil {
|
||||
var unknownUserErr user.UnknownUserError
|
||||
|
@ -154,6 +154,7 @@ func MakeUserNsManager(kl userNsPodsManager) (*UsernsManager, error) {
|
||||
}
|
||||
off := int(kubeletMappingID / userNsLength)
|
||||
len := int(kubeletMappingLen / userNsLength)
|
||||
klog.V(5).InfoS("User namespace manager mapping", "offset", off, "length", len, "idsPerPod", userNsLength)
|
||||
|
||||
m := UsernsManager{
|
||||
used: allocator.NewAllocationMap(len, "user namespaces"),
|
||||
|
@ -19,6 +19,8 @@ package node
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"os/exec"
|
||||
"strconv"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
@ -43,6 +45,10 @@ import (
|
||||
var (
|
||||
// non-root UID used in tests.
|
||||
nonRootTestUserID = int64(1000)
|
||||
|
||||
// kubelet user used for userns mapping.
|
||||
kubeletUserForUsernsMapping = "kubelet"
|
||||
getsubuidsBinary = "getsubids"
|
||||
)
|
||||
|
||||
var _ = SIGDescribe("Security Context", func() {
|
||||
@ -54,6 +60,8 @@ var _ = SIGDescribe("Security Context", func() {
|
||||
})
|
||||
|
||||
ginkgo.Context("When creating a pod with HostUsers", func() {
|
||||
e2eskipper.SkipIfNodeOSDistroIs("windows")
|
||||
|
||||
containerName := "userns-test"
|
||||
makePod := func(hostUsers bool) *v1.Pod {
|
||||
return &v1.Pod{
|
||||
@ -112,6 +120,74 @@ var _ = SIGDescribe("Security Context", func() {
|
||||
}
|
||||
})
|
||||
|
||||
f.It("must create the user namespace in the configured hostUID/hostGID range [LinuxOnly]", feature.UserNamespacesSupport, func(ctx context.Context) {
|
||||
// We need to check with the binary "getsubuids" the mappings for the kubelet.
|
||||
// If something is not present, we skip the test as the node wasn't configured to run this test.
|
||||
id, length, err := kubeletUsernsMappings(getsubuidsBinary)
|
||||
if err != nil {
|
||||
e2eskipper.Skipf("node is not setup for userns with kubelet mappings: %v", err)
|
||||
}
|
||||
|
||||
for i := 0; i < 4; i++ {
|
||||
// makePod(false) creates the pod with user namespace
|
||||
podClient := e2epod.PodClientNS(f, f.Namespace.Name)
|
||||
createdPod := podClient.Create(ctx, makePod(false))
|
||||
ginkgo.DeferCleanup(func(ctx context.Context) {
|
||||
ginkgo.By("delete the pods")
|
||||
podClient.DeleteSync(ctx, createdPod.Name, metav1.DeleteOptions{}, f.Timeouts.PodDelete)
|
||||
})
|
||||
getLogs := func(pod *v1.Pod) (string, error) {
|
||||
err := e2epod.WaitForPodSuccessInNamespaceTimeout(ctx, f.ClientSet, createdPod.Name, f.Namespace.Name, f.Timeouts.PodStart)
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
podStatus, err := podClient.Get(ctx, pod.Name, metav1.GetOptions{})
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
return e2epod.GetPodLogs(ctx, f.ClientSet, f.Namespace.Name, podStatus.Name, containerName)
|
||||
}
|
||||
|
||||
logs, err := getLogs(createdPod)
|
||||
framework.ExpectNoError(err)
|
||||
|
||||
// The hostUID is the second field in the /proc/self/uid_map file.
|
||||
hostMap := strings.Fields(logs)
|
||||
if len(hostMap) != 3 {
|
||||
framework.Failf("can't detect hostUID for container, is the format of /proc/self/uid_map correct?")
|
||||
}
|
||||
|
||||
tmp, err := strconv.ParseUint(hostMap[1], 10, 32)
|
||||
if err != nil {
|
||||
framework.Failf("can't convert hostUID to int: %v", err)
|
||||
}
|
||||
hostUID := uint32(tmp)
|
||||
|
||||
// Here we check the pod got a userns mapping within the range
|
||||
// configured for the kubelet.
|
||||
// To make sure the pod mapping doesn't fall within range by chance,
|
||||
// we do the following:
|
||||
// * The configured kubelet range as small as possible (enough to
|
||||
// fit 110 pods, the default of the kubelet) to minimize the chance
|
||||
// of this range being used "by chance" in the node configuration.
|
||||
// * We also run this in a loop, so it is less likely to get lucky
|
||||
// several times in a row.
|
||||
//
|
||||
// There are 65536 ranges possible and we configured the kubelet to
|
||||
// use 110 of them. The chances of this test passing by chance 4
|
||||
// times in a row and the kubelet not using only the configured
|
||||
// range are:
|
||||
//
|
||||
// (110/65536) ^ 4 = 4.73e-12. IOW, less than 1 in a trillion.
|
||||
//
|
||||
// Furthermore, the unit tests would also need to be buggy and not
|
||||
// detect the bug. We expect to catch off-by-one errors there.
|
||||
if hostUID < id || hostUID > id+length {
|
||||
framework.Failf("user namespace created outside of the configured range. Expected range: %v-%v, got: %v", id, id+length, hostUID)
|
||||
}
|
||||
}
|
||||
})
|
||||
|
||||
f.It("must not create the user namespace if set to true [LinuxOnly]", feature.UserNamespacesSupport, func(ctx context.Context) {
|
||||
// with hostUsers=true the pod must use the host user namespace
|
||||
pod := makePod(true)
|
||||
@ -124,8 +200,6 @@ var _ = SIGDescribe("Security Context", func() {
|
||||
})
|
||||
|
||||
f.It("should mount all volumes with proper permissions with hostUsers=false [LinuxOnly]", feature.UserNamespacesSupport, func(ctx context.Context) {
|
||||
// Create all volume types supported: configmap, secret, downwardAPI, projected.
|
||||
|
||||
// Create configmap.
|
||||
name := "userns-volumes-test-" + string(uuid.NewUUID())
|
||||
configMap := newConfigMap(f, name)
|
||||
@ -636,6 +710,8 @@ var _ = SIGDescribe("Security Context", func() {
|
||||
})
|
||||
|
||||
var _ = SIGDescribe("User Namespaces for Pod Security Standards [LinuxOnly]", func() {
|
||||
e2eskipper.SkipIfNodeOSDistroIs("windows")
|
||||
|
||||
f := framework.NewDefaultFramework("user-namespaces-pss-test")
|
||||
f.NamespacePodSecurityLevel = admissionapi.LevelRestricted
|
||||
|
||||
@ -683,3 +759,57 @@ func waitForFailure(ctx context.Context, f *framework.Framework, name string, ti
|
||||
},
|
||||
)).To(gomega.Succeed(), "wait for pod %q to fail", name)
|
||||
}
|
||||
|
||||
// parseGetSubIdsOutput parses the output from the `getsubids` tool, which is used to query subordinate user or group ID ranges for
|
||||
// a given user or group. getsubids produces a line for each mapping configured.
|
||||
// Here we expect that there is a single mapping, and the same values are used for the subordinate user and group ID ranges.
|
||||
// The output is something like:
|
||||
// $ getsubids kubelet
|
||||
// 0: kubelet 65536 2147483648
|
||||
// $ getsubids -g kubelet
|
||||
// 0: kubelet 65536 2147483648
|
||||
// XXX: this is a c&p from pkg/kubelet/kubelet_pods.go. It is simpler to c&p than to try to reuse it.
|
||||
func parseGetSubIdsOutput(input string) (uint32, uint32, error) {
|
||||
lines := strings.Split(strings.Trim(input, "\n"), "\n")
|
||||
if len(lines) != 1 {
|
||||
return 0, 0, fmt.Errorf("error parsing line %q: it must contain only one line", input)
|
||||
}
|
||||
|
||||
parts := strings.Fields(lines[0])
|
||||
if len(parts) != 4 {
|
||||
return 0, 0, fmt.Errorf("invalid line %q", input)
|
||||
}
|
||||
|
||||
// Parsing the numbers
|
||||
num1, err := strconv.ParseUint(parts[2], 10, 32)
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("error parsing line %q: %w", input, err)
|
||||
}
|
||||
|
||||
num2, err := strconv.ParseUint(parts[3], 10, 32)
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("error parsing line %q: %w", input, err)
|
||||
}
|
||||
|
||||
return uint32(num1), uint32(num2), nil
|
||||
}
|
||||
|
||||
func kubeletUsernsMappings(subuidBinary string) (uint32, uint32, error) {
|
||||
cmd, err := exec.LookPath(getsubuidsBinary)
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("getsubids binary not found in PATH")
|
||||
}
|
||||
outUids, err := exec.Command(cmd, kubeletUserForUsernsMapping).Output()
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("no additional uids for user %q: %w", kubeletUserForUsernsMapping, err)
|
||||
}
|
||||
outGids, err := exec.Command(cmd, "-g", kubeletUserForUsernsMapping).Output()
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("no additional gids for user %q", kubeletUserForUsernsMapping)
|
||||
}
|
||||
if string(outUids) != string(outGids) {
|
||||
return 0, 0, fmt.Errorf("mismatched subuids and subgids for user %q", kubeletUserForUsernsMapping)
|
||||
}
|
||||
|
||||
return parseGetSubIdsOutput(string(outUids))
|
||||
}
|
||||
|
Loading…
Reference in New Issue
Block a user